This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
And yet it was precisely a matter of status that created such discomfort between Hartmann and his mistress. (AB1)8 In (17) the ªrst thematic element marks conjunction, while the second contains a lexical tie, namely the use of a general word (event) that sums up the content of the preceding clause. Example (18) shows a juxtaposition of two conjunctive ties expressing diŸerent relations, namely additive (and) and adversative (yet). The example is typical in that ties marking additive conjunction frequently co-occur with other types of conjunctive ties, i.e. with adversative, causative and temporal ties. In the great majority of these examples, the additive conjunction is and (there is one example of or). This suggests that and may have a rather weakened or generalised meaning of addition/continuation, as was demonstrated also in the case of (13) above.
6.
Paragraph-initial multiple Themes
The beginning of a sentence is often seen as a strategically important position, where continuity as well as breaks in continuity can be marked (cf. Virtanen 1987: 55). We have already seen that multiple Themes tend to contain markers of cohesion, but it is also possible that they may be of service in introducing a new discourse topic. Longacre (1979: 118) claims that “there is a certain reluctance on the part of the speaker to plunge immediately into a topic. He wants to spend a sentence orienting himself and the audience to what he is going to talk about”. This ‘orienting’ section is termed an ‘ice breaker’ since it opens up for a new topic. The ice breaker function may be ªlled by “something that indicates time, place, or circumstances or gives a broad hint of what is to come in the body of the paragraph” (ibid.). This deªnition has obvious similarities with the deªnition of the function of Theme. It is thus of interest to study the extent to which multiple Themes in paragraph-initial position launch a new topic or mark continuity with the preceding paragraph(s). Example (19) shows a paragraph-initial multiple Theme marking a shift in the discourse, to a description of the occasions when ‘friends from the old days’ were present. (19) The washing-up done, the shadowy girl would come in with a tray of tea. “Thank you, dear,” Aunt would say, and would put extra sugar from her ration into Fibich’s cup. Such kindnesses he remembered.
Occasionally friends from the old days would drop in, fat men in stained tweed suits, women with unadorned faces. (AB1) However, it seems that paragraph-initial multiple Themes mark continuity more
77
78
Hilde Hasselgård
often than they mark a major textual shift. Table 8 shows the composition of paragraph-initial multiple Themes as regards the types of cohesive ties they contain. The percentage of non-cohesive Themes is greater in this position than in elsewhere: 18 per cent vs. 5 per cent. It turns out, however, that even in paragraph-initial position, multiple Themes tend to express cohesive relations, and hence some degree of continuity. Only 6 out of the 34 contain no cohesive ties. Although not too much should be concluded from such small ªgures, the ªndings indicate that it is more common for multiple Themes to mark continuity than to mark breaks. Table 8. Cohesive ties in paragraph-initial multiple Themes Theme 1
Theme 2
Theme 3
N
conjunction conjunction conjunction conjunction conjunction conjunction lexical reference reference reference non-cohesive non-cohesive
conjunction conjunction conjunction lexical reference non-cohesive non-cohesive lexical reference non-cohesive reference non-cohesive
non-cohesive lexical reference
3 1 1 1 7 3 1 1 1 4 5 6 34
Furthermore, a multiple Theme may well mark a minor shift and continuity at the same time. In (20), the conjunct ªnally places the clause message as a new stage in the same sequence of events. (20) She may have been a competent photographer: but she was so shortsighted, and so clumsy when handling the porcelains that Utz had to sit on the edge of his bed, numbly waiting for the crash. He begged to be allowed to position each piece in front of the camera. He was told it was none of his business.
Finally, when the young woman dropped, and smashed the head oŸ, a ªgure of Watteau’s Gilles, he lost his temper. (BC1) The deªnite article in the young woman refers to a participant in the preceding paragraph, and ªgure collocates with the porcelains just referred to, as well as with other words in the preceding context. This paragraph-initial multiple Theme thus marks a minor move in the discourse ¶ow by combining a paragraph boundary with tight cohesive linking.
The role of multiple themes in cohesion
In a couple of examples a paragraph-initial, non-cohesive Theme sets a new topic for the paragraph, as in (21), which describes the wildlife that can be sighted in a particular place. Four out of ªve successive Themes are derived from the ‘wildlife’ hypertheme (cf. Daneš 1974: 119). (21)
Sometimes fallow deer can be seen among the trees. Red squirrels have been sighted. The blackcock is a rarity, the Dartford warbler common, and hen harriers are winter visitors. In late spring, when the rhododendrons come out, the long vistas are rosy pink under a green mist of unfolding beech leaves. The nightingale sings. (RR1) Similarly, a sequence of thematic adjunct adverbials can be used to signal a shift in the discourse, as in (22). (22) He had collaborated. He had given information: a trickle of information as to the whereabouts of certain works of art-information available to anyone who knew how to use an art library. By doing so, he had been able to protect, even to hide, a number of his Jewish friends: among them the celebrated Hebraist, Zikmund Kraus. What, after all, was the value of a Titian or a Tiepolo if one human life could be saved?
As for the Communists, once he realised the Beneš Government would fall, he began to curry favour with the bosses-to-be. (BC1) In this paragraph the narrative takes a leap forward to the time when Czechoslovakia was under Communist regime. This shift is marked by the paragraph-initial adverbial. The word Communists is mentioned for the ªrst time, though it represents a contrast with the German Nazis implicitly referred to in the preceding paragraph. However, provided the sentence is not text- or chapter-initial, at least one of the adjuncts in a paragraph-initial sequence tends to contain a cohesive tie. In (22) this is provided by the use of the pronoun he in the adverbial subclause. Example (23), on the other hand, is chapter-initial, and the thematic sequence of adjuncts serves to provides a setting for what follows. (23)
In the summer of 1967 — a year before the Soviet tanks overran Czechoslovakia — I went to Prague for a week of historical research. (BC1) This example shows the ‘ice breaker’ function described by Longacre (1979: 118). At the same time it clearly serves the important thematic function of providing a local context within which the rest of the clause, and indeed the rest of the paragraph, should be interpreted.
79
80
Hilde Hasselgård
7.
Some contrastive observations
The texts used for the present investigation are available with translation into Norwegian in the ENPC (see Section 4), while three of them are also available in German in the OMC. A study of translations is revealing, as the translations represent an interpretation of the source text. In the case of multiple Themes, it is clear from the material that the translators have made several changes. This is to a great extent caused by the verb-second constraint in Norwegian and German, limiting the possible number of constituents in pre-verbal position. In fact, multiple Themes are generally less common in Norwegian and German than in English; cf. Hasselgård (2000: 18) and Teich (2001: 203). Thus many of the multiple Themes in the English material correspond to simple Themes in the translations. These simpliªed Themes provide interesting data for investigating which part of the Theme translators ªnd most indispensable for the textual structure of the sentence. The translations of (24) seem to be typical to the extent that the simpliªed Themes in both Norwegian and German show a clear tendency for textual and interpersonal Themes to be either omitted or moved to a later position, while the topical Theme of the original is preserved in the translation.9 (24) Several of my friends at the Sorbonne went oŸ to Austria to help the refugees. In fact, from all over the world students swarmed to Vienna. (ABR1) Norwegian: Fra alle verdens kanter strømmet det faktisk studenter til Wien. (Lit: From all the-world’s ends streamed there in-fact students to Vienna.) German: Aus der ganzen Welt kamen Studenten in Scharen nach Wien. (Lit: From the whole world came students in crowds to Vienna.) We may also ªnd more thorough restructuring, as in (25); cf. also Table 9. The Norwegian translation has expressed the meaning of the ªrst adverbial by means of a verb phrase, while the German translation has moved the second adverbial to clause-medial position. (25) Sometimes, in the evenings, he would forfeit his place in front of the electric ªre and disappear to be sick. (AB1) Norwegian: Om kvelden hendte det at han gav slipp på sin plass foran den elektriske ovnen og forsvant ut for å kaste opp. (Lit: In the-evening happened it that he gave up his place…) German: Manchmal verzichtete er am Abend auf seinen Platz am elektrischen Feuer und verschwand, um sich zu übergeben (Lit: Often forfeited he in-the evening his place…)
The role of multiple themes in cohesion
In Hasselgård (2000: 25) it was found that translators generally try to preserve the topical Theme of the original when they translate. This is an even clearer tendency in translations from English to Norwegian than in translation from English to German, where the ªrst Thematic element of the source sentence is sometimes preserved at the cost of the topical Theme (ibid: 29). This tendency is illustrated by (26). (26) Marta, whom he had taught to make a sou§é, asked what time the guest would arrive. He stood up. He straightened his tie. Then, without a hint of condescension, he pressed her calloused hand to his lips. (BC1) Norwegian: Uten snev av nedlatenhet trykket han den barkede hånden hennes mot leppene. (Lit: Without hint of condescension pressed he…) German: Dann drückte er, ohne eine Spur der Herablassung, ihre schwielige Hand an seine Lippen. (Lit: Then pressed he, without a hint the[+gen.] condescension…) Here both the Norwegian and the German translator have split up the original sequence of adverbials in order to put the verb in second position. The Norwegian translation has retained the experiential Theme and omitted the textual one, leaving the temporal sequence implicit. The German translation has retained the marker of temporal conjunction in initial position, and in consequence has had to move the other adverbial into rhematic position. Thus, while the Norwegian translation has given priority to the experiential Theme in this case, the German one has opted for the explicit marking of cohesion. Both Norwegian and German seem to rely somewhat less on conjuncts than English, and use either conjunctions or no overt conjunctive relation to mark certain cohesive relations.10 The use of conjunctions instead of conjuncts of course allows the translation to retain the pattern of the original Theme, as seen in (27). (27) Yet in his mind he never entirely left the ¶at in Compayne Gardens, where he had stayed for ªve years after Aunt Marie’s death, until Hartmann, who was the ªrst to marry, found the two homes where they could be as nearly together as was compatible with their married state. (AB1) Norwegian: Men i sin sjel forlot han aldri helt leiligheten i Compayne Gardens… German: Aber im Geist verließ er nie ganz die Wohnung in den Compayne Gardens… In contrast to English, Norwegian and German have a set of modal particles. In the corpus, these sometimes correspond to modal disjuncts in English, as seen in (28).11 The modal particle jo/ja functions mainly as a marker of given information, which makes it an adequate rendering of of course. Modal particles are placed
81
82
Hilde Hasselgård
immediately after the ªnite verb (or after the subject in clauses with inversion), which usually means that they have to occur outside the Theme. Replacing an English thematic disjunct by a modal particle in the Norwegian or German translation may be a good way of removing the adverbial from thematic position while keeping the modal meaning within the mood section of the clause. (28) Of course you haven’t been here long, but you’ll have heard of Davina Flory? (RR1) Norwegian: Du har jo ikke vært her så lenge, men du har i hvert fall sikkert hørt om Davina Flory? (Lit: You have [modal] not been here so long…) German: Sie sind ja noch nicht lange hier, aber von Davina Flory haben Sie wohl schon gehört, nicht? (Lit: You are [modal] still not long here…) A slightly higher percentage of multiple Themes are simpliªed in the Norwegian than in the German translations: 32 per cent vs. 26 per cent. Table 9 shows the most common syntactic changes that are made in the simpliªcation process. The most frequent change in both the Norwegian and the German translations is that the interpersonal Themes are either deferred to rhematic position or omitted, as in (24). In other words, they tend to lose the competition for thematic position. However, textual elements are also sometimes moved out of thematic position or omitted, e.g. in the Norwegian translation of (26). Altenberg (1998: 124, 130) presents similar ªndings for Swedish, i.e. initial conjuncts in English often correspond to conjuncts in medial or end position in Swedish. It seems, along the lines of Altenberg’s conclusion (ibid.: 139), that textual elements tend to yield to experiential elements in the competition for thematic prominence in a V2 language. For the purposes of the present investigation we may note that translators tend to regard topicality as a more important function of Theme than the kinds of relationships expressed by disjuncts and conjuncts. Table 9. Changes made in thematic structure where the Norwegian and German translations have simple Themes Interpersonal Theme moved to rhematic position Interpersonal Theme omitted Textual Theme moved to rhematic position Textual Theme omitted Experiential Theme moved to rhematic position Textual Theme → Experiential Theme Conjunct → conjunction Other kinds of restructuring TOTAL
Norwegian
German
16 0 8 8 5 5 8 8 58
9 2 7 8 8 1 1 11 47
The role of multiple themes in cohesion
8.
Summary and conclusions
A major ªnding of the present study is that the great majority of multiple Themes in English — over 90 per cent — contain at least one cohesive tie. This suggests strongly that an important function of multiple Themes is to latch a sentence explicitly on to the preceding context. Well over half of the multiple Themes in the material consist of a combination of one element that functions as a cohesive tie and one or two non-cohesive elements. The present material has thus supported the hypothesis that multiple Themes can bring a non-cohesive element into thematic position by coupling it with one that secures cohesion (cf. Section 2). In contrast to simple Themes, which consist of a single, experiential element, a multiple Theme can mark two or more cohesive relations at the same time. Hence, the other hypotheses concerned multiple Themes with two or more cohesive ties. Such Themes were found to generally express two diŸerent kinds of cohesive relation. The most common combination was conjunction and reference. Although we rarely see two ties of the same type within the same Theme, there were examples of two diŸerent conjunctive relations or two elements marking reference within the same Theme. In cases of two diŸerent conjunctive relations, one of them is usually additive (the and-relation). Paragraph-initial multiple Themes were looked at speciªcally, because it was assumed that they would typically signal a shift in the discourse. Although this function was attested in the material, the majority of paragraph-initial multiple Themes marked continuity as well. About half of them contained both cohesive and non-cohesive elements, and thus tended to signal a minor move in the discourse rather than major break with the preceding context. As regards the kind of cohesive link that is created by a multiple Theme, we found, in descending order of frequency: reference, conjunction, and lexical cohesion. The distribution of cohesive ties in multiple Themes is diŸerent from their general distribution over the whole clause. The multiple Themes contain a much higher proportion of conjunctive ties, most likely because such ties are typically found in clause-initial position in English. Here English diŸers from Norwegian and German, which have a tendency to place conjunctive adverbials in post-verbal, and hence rhematic, position. Lexical cohesion seems to be greatly underrepresented in multiple Themes as compared to the general distribution of such ties. Furthermore, ellipsis and substitution are not attested at all in the multiple Themes in the material. We have also seen that the components of multiple Themes diŸer as to their likelihood of being cohesive. Textual Themes are nearly always cohesive, to the extent that the marking of cohesive relations can be said to be their main function. About two thirds of experiential Themes are cohesive; they generally express refer-
83
84
Hilde Hasselgård
ence or lexical cohesion, both of which are typical properties of noun phrases. Interpersonal Themes are hardly ever cohesive. They are, however, important in signalling that the clause should be interpreted in a particular light, e.g. as the speaker’s opinion of or hypothesis about a state of aŸairs. The study of translations of English multiple Themes into Norwegian and German revealed a few important diŸerences between English on the one hand and the verb-second languages Norwegian and German on the other. The ªrst is directly related to the verb-second constraint, which limits the number of elements that can be accommodated before the subject-verb nexus. In particular, Norwegian and German have less scope for sequences of adverbials in initial position, which is of some consequence for the structure and use of multiple Themes. By contrast, all kinds of adverbials can easily be accommodated in medial position in both of these languages. Thus we see that some of the meanings that are expressed in multiple Themes in English, particularly interpersonal meanings and conjunctive relations, are expressed post-verbally in the translations. Conjunctive relations also seem to be left implicit slightly more often in German and Norwegian than in English, and if they are expressed, they tend to be realized by conjunctions more often than by conjunct adverbials. What is perhaps missing from the present study is an examination of information structure, which is the third major component of texture; cf. Halliday (1994: 334). In clauses with unmarked information structure the Theme coincides with given information (ibid.: 301). Markers of given information are to a great extent the same as those that mark the cohesive relation of reference, e.g. the use of pronouns or deªnite and demonstrative determiners. There is thus an evident interplay between information structure and Theme on the one hand, and information structure and cohesion on the other, and a future study might beneªt from taking all three aspects of texture into account. Nevertheless, the ªndings of the present study corroborate the notion that much textual work is done at the beginning of a clause. They also suggest that English Themes are Janus-faced, in the sense that they tend to signal a relation with the preceding context at the same time as they lay a foundation for a new clause.
Notes 1. Sequences of initial adjunct adverbials are treated as thematic also by Matthiessen (1992: 50). 2. In this and following examples, the multiple Theme under discussion has been highlighted. 3. For a description of the corpora, see www.hf.uio.no/iba/prosjekt/ and www.hf.uio.no/ german/sprik/english/corpus.shtml.
The role of multiple themes in cohesion
4. This gives a frequency of 3.7 multiple Themes per 1,000 words. 5. The total number of ties is slightly higher than the total number of Themes, because a few ties have been marked for more than one type of cohesive relation. 6. Obviously, an experiential theme also has implications for the interpersonal and textual metafunctions, but one may claim its primary function is reference to a participant, circumstance or process. Interpersonal and textual Themes, on the other hand, do not have experiential functions. 7. Thanks to Eija Ventola (pers. comm.) for pointing this out to me. 8. The tag
signals a paragraph boundary. 9. Norwegian and German translations. 10. Incidentally, the same tendency was found by Teich (2001: 203), who investigated translations between English and German. 11. The correspondence between modal particles and disjunct adverbials can also be seen in translations from Norwegian into English, cf. Hasselgård (2000: 33).
References Altenberg, Bengt 1998 “Connectors and sentence openings in English and Swedish”. In Corpora and Cross-linguistic Research, S. Johansson and S. Oksefjell (eds), 115–144. Amsterdam: Rodopi. Biber, Douglas, Johansson, Stig, Leech, GeoŸrey, Conrad, Susan, and Finegan, Edward 1999 Longman Grammar of Spoken and Written English. London: Longman. Cloran, Carmel 1995 “Deªning and relating text segments: Subject and Theme in discourse”. In R. Hasan and P. H. Fries (eds). 361–403. Daneš, František 1974 “Functional Sentence Perspective and the organization of the text”. In Papers on Functional Sentence Perspective, F. Daneš (ed.), 106–128. Prague: Academia. Enkvist, Nils Erik 1976 “Notes on valency, semantic scope and thematic perspective as parameters of adverbial placement in English”. In Reports on Text Linguistics: Approaches to Word Order, N. E. Enkvist and V. Kohonen (eds), 51–73. Åbo: Åbo Akademi. Fries, Peter H. 1995 “Themes, methods of development, and texts”. In R. Hasan and P. H. Fries (eds). 317–359. Halliday, M. A. K. 1994 An Introduction to Functional Grammar. London: Edward Arnold.
85
86
Hilde Hasselgård
Halliday, M. A. K. and Hasan, Ruqaiya 1976 Cohesion in English. London: Longman. 1985 Language, Context, and Text: Aspects of Language in a Social-semiotic Perspective. Geelong: Deakin University Press. Hasan, Ruqaiya and Fries, Peter H. (eds) 1995 On Subject and Theme. Amsterdam: John Benjamins. Hasselgård, Hilde 1996 Where and When: Positional and Functional Conventions for Sequences of Time and Space Adverbials in Present-Day English. [Acta Humaniora] Oslo: Scandinavian University Press. 2000 “English multiple themes in translation”. In Contrastive Studies in Syntax. Copenhagen Studies in Language 25, A. Klinge (ed.). 11–38. Copenhagen: Samfundslitteratur. Longacre, Robert 1979 “The paragraph as a grammatical unit”. In Syntax and Semantics, vol. 12. Discourse and Syntax, T. Givón (ed.). 115–134. New York: Academic Press. Matthiessen, Christian 1992 “Interpreting the textual metafunction”. In Advances in Systemic Linguistics: Recent Theory and Practice, M. Davies and L. Ravelli (eds), 37–81. London: Francis Pinter. 1995 “Theme as an enabling resource in ideational ‘knowledge’ construction”. In Ghadessy, Moshen (ed.). Thematic Development in English Texts. 20–54. London: Pinter. Matthiessen, Christian and Halliday, M. A. K. 1997 Systemic Functional Grammar: A First Step into the Theory. Ms, accessed from ‘Virtual Library’ at
The role of multiple themes in cohesion
Sources of data (selections from the ENPC/OMC: http://www.hf.uio.no/iba/prosjekt/) ABR: Brink, André, The Wall of the Plague, Faber and Faber Ltd, London, 1984. Malde, Per. Pestens mur, H. Aschehoug & Co, Oslo, 1984. Peterich, Werner, Die Pestmauer. Kiepenheuer & Witsch, Köln, 1984. AB1: Brookner, Anita. Latecomers, Jonathan Cape, London, 1988. Jahr, Mette-Cathrine, Etternølere, Gyldendal Norsk Forlag AS, Oslo, 1990. Walter, Edith, Nachzügler. Zsolnay, Wien, 1991. BC1: Chatwin, Bruce, Utz, Jonathan Cape, London, 1988. GreiŸ, Aud, Utz, H. Aschehoug & Co, Oslo, 1989. Kamp, Anna, Utz. Fischer Taschenbuch Verlag, Frankfurt am Main, 1996. RR1: Rendell, Ruth. Kissing the Gunner’s Daughter, Hutchinson, London, 1992. Tønnesson, Birgit. Brent barn, H. Aschehoug & Co, Oslo, 1993.
87
Cohesion and collaboration Patterns of cohesion in spoken and written dialogue Sanna-Kaisa Tanskanen University of Turku
1.
Introduction
The purpose of this paper is to discuss cohesion patterns in spoken and written dialogue, more speciªcally in face-to-face conversation and e-mail mailing-list messages. The paper concentrates on the communicators’ use of explicit cohesive markers, which are considered important in contributing to coherence, but not necessary nor su¹cient for guaranteeing it. Cohesion and coherence are thus viewed as independent but intertwined: cohesion refers to grammatical and lexical elements on the surface of a text which can form connections between parts of the text, while coherence resides not in the text, but is the outcome of a dialogue between the text and its listener or reader.1 The present study rests on the conviction that cohesion contributes to coherence, i.e. cohesion is one of the ways of signalling coherence in texts. Although the focus is on these linguistic signals of coherence, the paper also highlights cognitive aspects underlying the construction of coherence. It is argued that cohesion can be viewed as a collaborative device used by communicators as a signal of their attempt to successfully interact with their fellow communicators.
2.
Cohesion, coherence and collaboration
Several recent works on language in communication point out that it is not a su¹cient or satisfactory solution to look at the communication process solely or mainly from the perspective of the producer. The two quotations below, one from Brown and the other from Clark, point in this direction:
90
Sanna-Kaisa Tanskanen
An account of communication which assumes that only the speaker’s intentions need to be taken into account is as inadequate as one which assumes that speaker and listener will share common goals and a common context (Brown 1995: 24). Language use is really a form of joint action, …carried out by an ensemble of people acting in coordination with each other. …Doing things with language is diŸerent from the sum of a speaker speaking and a listener listening. It is the joint action that emerges when speakers and listeners — or writers and readers — perform their individual actions in coordination, as ensembles (Clark 1996: 3).
Both Brown and Clark emphasise the fact that communication requires two parties: it is therefore basically a collaborative process. The active role in communication of all the communicators constitutes one of the main arguments of the proponents of the collaborative view of communication. They assert that communicators actively collaborate to ensure that understanding takes place (Anderson 1995; Brown 1995; Bublitz & Lenk 1999; Clark 1996:passim; Clark & Wilkes-Gibbs 1986; Coates 1995; Fox Tree 1999; Linell 1998: esp. Ch. 11 & 13; Schober & Clark 1989; Traxler & Gernsbacher 1995). Linell (1998) proposes that the models which emphasise the interactive and collaborative nature of communication could be grouped under the general heading of dialogism. Although varied in their approaches, dialogic models view communication as a collaborative accomplishment, whereas monologic models would rather stress the role of the (current) speaker or writer in the production of discourse. Monologists see communication as “a process between individuals”, while for dialogists, communication is about “individuals in dialogue with partners and contexts” (Linell 1998: 8). The collaboration between communicators is clearly visible in conversation, in which participants actively cooperate to achieve coherence. Collaboration can be realised for example as feedback between participants in the form of completions, clarifying questions or other types of acknowledging that the participants have understood what their fellow communicators were saying (Anderson et al. 1997; Brown 1995:passim; Coates 1995; Traxler & Gernsbacher 1995; Wilkes-Gibbs 1995). However, such signals are by no means necessary for successful collaboration. In fact, one might argue that a continuous ¶ow of communication is in itself a proof of collaboration and that no overt signals of cooperation, that is overt in the sense that their purpose clearly is to provide feedback, are needed. In addition, there are covert signals which have not yet received similar attention, but which may equally well contribute to collaboration. An example would be cohesive devices. These have not yet been analysed from a collaborative perspective, but they may proªtably be included in such analyses as suggested by Dickinson and Givón:
Cohesion and collaboration
During face-to-face communication, speakers-hearers are engaged simultaneously in a great number of tasks; most prominent among them are the management of cooperative interaction between the interlocutors, and the processing of coherent information-¶ow. It seems altogether reasonable that the two tasks are not totally divorced from each other (Dickinson & Givón 1997).
In conversation, therefore, the way in which interlocutors use cohesive devices can be a signal of their attempts to successfully interact with their fellow interlocutors (see also Hoey 2001: 39–41; Linell 1998: 182). The collaborative or dialogical view of communication can be regarded as somewhat idealistic, or as Linell (1998: 276) phrases it, it would be “foolish to argue simplistically that the world of human experience is characterized by a maximal degree of dialogicality”.2 Nevertheless, collaboration theory brings out the cognitive aspects of the communication process, while stressing the role of the communicators. It follows that collaborative or dialogical models therefore appear extremely suitable to show a successful interplay between cohesion, coherence and communicators, and they can consequently be used in the eŸort to try to understand the demands, but also the possibilities, of a communicative situation.
3.
Lexical cohesion relations
3.1 Word versus lexical unit For a long time, the orthographic word has been the unit of analysis in lexical studies (Sinclair 1998: 2; Weigand 1998: 39). In some studies, the term lexical item is also used about a single word, and larger units are referred to as multiple lexical items (Carter 1987: 7–8; Halliday 1989: 63; 1994: 330–331; Martin 1992: 290; Nyyssönen 1992). Let us consider the consequences of regarding the orthographic word as the basic unit of analysis in the context of a textual (cohesion) study. The units which enter into cohesive relations are clearly not orthographically restricted: phrasal verbs and idioms, as well as other “multiple lexical items” can and do take part in lexical cohesion. Martin (1992: 293, 326) remarks that the units which form cohesive relations can be realised by a single lexical item, i.e. a word, or by multiple lexical items; for the functioning of cohesion the number of items in a unit is irrelevant (see also Halliday 1994: 311). There are thus good arguments for postulating a larger basic unit than the word for the purpose of studying cohesion: if considered separately, words within a unit can enter into relations that are very diŸerent from those they enter into when regarded as a single unit (see also Martin 1992: 291). In other words, a single word can enter into diŸerent lexical relations from the larger unit of which it is a part.
91
92
Sanna-Kaisa Tanskanen
In the present study, it is argued that lexical cohesion relations are conditioned by their context; i.e. they are created, used and interpreted within the conªnes of their context. Given this hypothesis, it is natural that the deªnition of lexical units is also dependent on the text in which they are used. Following Nyyssönen (1992) this could perhaps be called a discourse-based approach to lexis, here applied speciªcally to lexical units (see also Hoey 1991: 220). Let us see what the context-based approach to cohesion means in practice by considering a couple of examples. It should ªrst be noted that many of the units recognised by the analysis are realised as single orthographic words, thus posing few problems for the analyst. However, a large number of the units are multi-word units. Phrasal verbs and idioms may be the most straight-forward examples of multiword units. Thus, when established is repeated as set up in a text, it is not di¹cult to justify their relation, although the former is realised as a single orthographic word and the latter as two. The same is true when it comes to idioms: if we were to divide an idiom into orthographic words, it would not only lose its meaning, but the parts could also be related to units that the idiom as a whole is unrelated to. It is clear that for an analysis of cohesion the meaning of the unit must be more important than its orthography, and this also applies to units which cannot be classiªed as phrasal verbs or idioms. For instance, it would be hazardous to analyse the unit cultural determinism as two distinct words. Cultural on its own, with its wide meaning, could enter into relations with a great number of other words, whereas when it is used together with determinism its meaning is deªned by this cooccurrence, and as a whole the unit is related to, for instance, other social theories discussed in the same text. Similarly, units such as social services, Standard English, the working people and out of fashion are analysed as single lexical units. Finally, let us consider an example which illustrates the importance of the meaning of the unit as a whole. The relevant extract is presented below (with some intervening text deleted, marked with …). The extract comes from a university lecture taken from the British National Corpus and the potential diŸerences between the written and spoken forms of the unit make it even more interesting: (1) Carbon dioxide is the most soluble of the gases because as it dissolves it doesn’t just go through a physical solution it goes through a chemical conversion … This system is a dynamic equilibrium. If you put more C O two into the system the concentrations of all of these go up… (BNC F8E) Because the text we see is a transcription of an audiotape, we cannot know whether the unit that the lecturer ªrst refers to as carbon dioxide and a while later as C O two was actually written like that in his lecture notes, because he may have simply written it as CO2. Regardless of which spoken form he uses, he is however talking
Cohesion and collaboration
about the same chemical substance. A cohesive analysis that fails to recognise a relation between carbon dioxide, C O two and CO2 would obviously be imperfect. The above discussion and examples have hopefully shown why an analytical approach based on orthography, i.e. using a single orthographic word as the unit of analysis, would be less than ideal for a study of cohesion. Instead, the analysis presented in this paper rests on a view of lexical units, produced, interpreted and deªned in their context of use.
3.2 Reiteration and collocation relations One of the starting points for the present analysis is the conviction that all cohesively meaningful relations have to be included in the analysis. Consequently, the model of analysis includes both reiteration and collocation relations (see Halliday & Hasan 1976; Halliday 1994; Martin 1992; cf. Hasan 1984). In the following, I list the devices together with an example of each category (the classiªcation is discussed in greater detail and more examples are given in Tanskanen 2000).3 Let us start with the reiteration relations. Repetition is without doubt the most straightforward relation in the present study. It can be divided into simple repetition and complex repetition. Simple repetition occurs when a unit is repeated either in an identical form or with no other than a simple grammatical change, e.g. singular – plural, present tense – past tense. Complex repetition involves a more substantial change: the units may be identical but serve diŸerent grammatical functions, or they may not be identical but share a lexical morpheme. 1. Simple repetition A: oh are you playing the recorder too b: I play the recorder too and I ªnd this quite amusing…(LLC S. S.14) 2. Complex repetition And cultural determinism is the idea that (…) they interpreted Freud as if he too were a cultural determinist (BNC HUH) Substitution is the third subcategory of reiteration relations. This is a category that has traditionally been treated under grammatical cohesion (like pronoun repetition), but because substitution functions in a way very similar to lexical repetitions, it is included in the analysis (see also Hoey 1991: 73; Hasan 1984). 3. Substitution …such a decision opens up the “perfect” way for any racist employer or group to exclude blacks. All they would have to do would be to say that… (mailing list) The fourth subcategory of reiteration is equivalence. Following McCarthy (1988),
93
94
Sanna-Kaisa Tanskanen
the term equivalence is used to refer to the relation more commonly referred to as synonymy, the reason being that synonymy would fail to make explicit the discourse-speciªcity of the lexical relation. In other words, the analysis does not start from a ready-made classiªcation which would tell us which relations are possible; it starts from a text and tries to establish which units are related in that particular text.4 The diŸerence between a lexical-semantic and a discourse-speciªc approach can be illustrated by the distinction made by McCarthy (1988): a semantic analysis can show the “meaning potential” of units, whereas a discourse-speciªc approach is concerned with the “communicative potential” of units.5 The two approaches are thus clearly distinct but not irreconcilable; McCarthy observes that a discoursespeciªc analysis need neither undermine nor conªrm the analysis of a semanticist; it is a statement about use, which is diŸerent in kind from statements made in a decontextualized, structural description of the lexicon (McCarthy 1988).
Instead of applying a classiªcation used in lexical semantics, other labels for the categories of lexical relation will be used re¶ecting the fact that the communicative potential of lexical units is appreciated and that their meaning is considered as constructed and controlled by the context in which they appear. The discourse-speciªc approach to lexical relations explains, for example, that the language user may use a unit which is equivalent with another unit, although the units may not be absolutely synonymous. By using a term like equivalence, attention can be drawn to the fact that the justiªcation and explanation for a relation between lexical units can and should be sought for in the text in which the units occur. Equivalence is illustrated in 4: 4. Equivalence …and the things that distinguished the Nazi extermination of the Jews from other forms of oppression. (…) I also told him that it was an issue that aŸected me deeply, and that my extended family had lost many people to the Nazi slaughter. (mailing list) The ªfth subcategory is generalisation, which covers the relation between a unit and a more general unit. 5. Generalisation …to reduce dependence on imported oil (…) In particular, they have made considerable progress in freeing internal markets for energy products. (SEC C01) Speciªcation, the sixth subcategory, is the opposite of generalisation: it refers to the relation between a unit and a more speciªc unit.
Cohesion and collaboration
6. Speciªcation Dialects which, however widely used in the present, do not access needed information are inferior with respect to those purposes which require that information than those that do not. When I could aquire the sum and total of western science in “Ebonican”, then, and only then… (mailing list) The next subcategory of reiteration is co-speciªcation, which includes the relation between two units which have a common general unit (World English in 7). 7. Co-speciªcation C: …one branch of it is Indian English because… b: well I’m sure another branch is South African English (LLC S.2.14) The ªnal subcategory of reiteration is contrast, which refers to the relation between a unit and another unit which has an opposite meaning. Again, the units that are considered to be related by contrast need not be strictly antonymous in the lexical semantic sense. What is important is that the units in question are used in a contrasting way in a particular text. 8. Contrast …a fashion of writing which went dramatically out of fashion immediately after WWI. So, at the time it was published most readers would have regarded it as completely up to date in its style… (BNC HUH) Having introduced all the reiteration categories, it is now time to discuss collocation relations. This category has been notoriously di¹cult to deªne, with the result that it has often been excluded from analyses (see e.g. Hasan 1984). Halliday and Hasan (1976: 284) say that collocation is achieved through the association created by habitually co-occurring lexical items. The items occur in similar environments because they describe things or happenings that occur in similar situations. Although this deªnition is vague, it will be the basis for the analysis. However, starting from this basis, it is necessary to try to deªne collocation relations in order to make the model of analysis replicable and objective. Ordered set is the ªrst of the three collocation categories in the classiªcation. It is perhaps the clearest of the categories and closest to the more systematic reiteration relations described above. The category includes members of ordered sets of lexical units, for example, colours, numbers, months, days of the week and the like. 9. Ordered set Judy: Monday? Doreen: First thing, ªrst thing. Mm. And come back about nine o’clock the Saturday night. (BNC KCR)
95
96
Sanna-Kaisa Tanskanen
The other relations discussed by Halliday and Hasan (1976) under collocation are more problematic to deªne. They are by deªnition nonsystematic, based only on an association between items, thus resisting systematic classiªcations and deªnitions. Consequently, it is impossible to construct watertight rules or models which would always tell us which units are related and which are not.There are, however, some previous studies which may help us understand and classify these complex relations. Martin (1992: 309–325) presents a redeªnition of Halliday and Hasan’s collocation category. In Martin’s model, nuclear relations re¶ect the ways in which “actions, people, places, things and qualities conªgure as activities”. As an example of such relations Martin mentions, among others, serve – ace. It appears that in the material as well there are pairs such as cyphers – decode or meals – eat or driving – the same car in which the relation between the units is based on an activity: you can decode cyphers, eat meals and drive cars. In classifying such units, it may thus be helpful to think of the association between the units as resulting from such a relation. This category will consequently be called activity-related collocation. 10. Activity-related collocation C: well I expect you don’t need cyphers during… if by that you mean people who e: people who can decode yeah (LLC S.2.3) However, even after recognising all the cohesive pairs based on the relations discussed above, there still remain relations which cannot be classiªed as ordered set or activity-related collocation. The relation in the example below (Holocaust – Nazi point of view) illustrates the third collocation category, elaborative collocation. 11. Elaborative collocation …on the topic of the ªring of Christina JeŸrey as US House of Representatives historian. The reason she was ªred was because she complained in 1986 that a proposed Jr. High curriculum on the Holocaust was not balanced or objective because it did not include the Nazi point of view… (mailing list) This category describes pairs whose relation is impossible to deªne in a more speciªc way than stating that the units can elaborate or expand on the same topic.6
4.
The material: spoken and written dialogue
As noted above, conversations show participants visibly collaborating, and for a study of collaborative coherence conversations thus form a natural starting point. The present study analyses two diŸerent conversations. Conversation 1, S.2.14. from the London-Lund Corpus of Spoken English, recorded in 1976, is a polyadic
Cohesion and collaboration
dialogue with three speakers. Conversation 2, J06 from the Lancaster IBM Spoken English Corpus, is a dyadic dialogue recorded in 1985. E-mail mailing-list messages have been described by DuBartell (1995) as “written, yet informal, spontaneous and conversational”. It is the attempt to at least partly imitate spoken dialogue through a written medium that makes mailing-list messages interesting material for an exploration of the collaborative aspects of cohesion and coherence. In the present paper, an e-mail mailing-list discussion on student bigotry is analysed and compared to the face-to-face conversations. The discussion took place on the WMST-List (Women’s Studies List) in February 1995. Using the term dialogue for mailing-list language may not be standard practice, but so many features typical of dialogue are nevertheless evident in e-mail messages that the term does not seem inappropriate. Linell’s deªnitions of monologue and dialogue are helpful here, since they are broad enough to subsume diŸerent kinds of text. In monologue, “only one person is, at least at face value, active as speaker or author”, whilst dialogue refers to “any dyadic or polyadic interaction between individuals who are mutually co-present to each other and who interact through language” (Linell 1998: 9). Linell mentions telephone conversations and electronic real-time interactions as special cases of dialogue, but I would like to extend it to cover mailing-list discussions as well. Although mailing-list messages are not produced synchronously and the communicators are not co-present, the messages deªnitely have an element of interactivity: with the help of an electronic medium, the communicators discuss a topic, express opinions and exchange ideas, albeit usually at a slower speed than in face-to-face or telephone conversations. Furthermore, the dialogicality of email messages can be apparent even on the surface of the messages. There are sequences in e-mail messages which clearly try to simulate conversation, i.e. sequences in which the current writer copies text from an earlier message and replies to it (commonly referred to as quoting or text-copying). This point will be returned to in the analysis.
5.
Patterns of cohesion in spoken dialogue
Let us start the analysis with spoken dialogue, i.e. the conversations. Some quantitative information will ªrst be provided on the occurrences of lexical cohesion relations in the conversations. Although reducing textual features to numbers may not always be an appropriate practice, in the case of reiteration and collocation relations this can undoubtedly help us see the most important general tendencies. This information might be harder to deduce from extracts from the texts, which will be considered here as well.
97
98
Sanna-Kaisa Tanskanen
It is clear from Table 1 that reiteration relations are far more frequent than collocation relations in the conversations. In Conversation 1, there are 109 reiteration pairs and nine collocation pairs per 1000 words, and in Conversation 2, 143 reiteration pairs and 15 collocation pairs. Although the number of collocation pairs is small compared to the number of reiteration pairs, I would like to argue that it is by no means unimportant. The exclusion of collocation relations from the analysis on the basis of the di¹culties connected with recognising them could indeed mean that an important aspect of lexical cohesion would remain unacknowledged. Table 1. Pairs of reiteration and collocation (per 1000 words) in the conversations7
Reiteration Collocation TOTAL
Conversation 1 (polyadic)
Conversation 2 (dyadic)
109 9 118
143 15 158
The diŸerence between Conversations 1 and 2, the former a polyadic and the latter a dyadic conversation, appears to be re¶ected in the number of cohesive pairs. Conversation 1, the three-party conversation has fewer pairs than Conversation 2, the two-party one; in terms of the overall number of pairs, the diŸerence is quite striking. The potential causes of this diŸerence are discussed later, since making any conclusions naturally requires more detailed information on the actual use of the cohesive pairs in the conversations. Yet, it is interesting that the number of cohesive pairs seems to diŸerentiate between the two types of conversation quite clearly. Let us next brie¶y examine the distribution of cohesive pairs for the categories of reiteration and collocation. Table 4 (Section 6) shows the normalised frequencies of the eight reiteration categories and the three collocation categories in the material analysed. The largest category by far in the conversations is simple repetition. Substitution and co-speciªcation come next, and pairs related by generalisation and speciªcation are also represented. Conversation 2 shows equivalence relations as well. Complex repetition is infrequent in the conversations, and so are, perhaps less surprisingly, the collocation categories of ordered set and activity-related collocation. Pairs of elaborative collocation are thus almost solely responsible for the occurrences of collocation relations. There are no pairs related by contrast. Let us next consider extracts from both conversations. Example (2) is from the three-party conversation and shows both reiteration and collocation relations at work.8 (2) A: …and having toyed with places like Malta and Cyprus he thought perhaps he would try South Africa
Cohesion and collaboration
C: yeah A: only I think he has he has a cousin who was a year or two in Cape Town and spoke warmly of Cape Town C: m A: so he went and tried Barry was just saying why did he get bitten with South Africa C: yes why choose South Africa indeed b: m A: well as I say because Vidor wanted a place with a warm climate where people spoke English so that he could earn some money b: can he really earn some money there… This example only shows reiteration relations: Malta, Cyprus and South Africa are ªrst related in terms of co-speciªcation; South Africa is related to Cape Town by speciªcation, and Cape Town is repeated. South Africa then appears as a generalisation of Cape Town and is repeated. Earn is repeated, as well as some money. He is repeated several times, until eventually Vidor replaces the pronoun only to be replaced by he a little later. As this example shows, substitution works in both directions: from noun to pronoun and pronoun to noun. In example (2), speaker A also refers to herself by I. It is interesting to note in this context that ªrst-person (and second-person) pronoun repetitions have sometimes been considered as problematic in cohesion analyses, because they refer to persons external to the text and thus their role in the cohesive constitution of a text may not be quite clear (cf. Hoey 1991: 71 and passim). It should be added that in several analyses the problem has been avoided by not including pronoun repetition, but this is hardly a commendable solution. If we consider the context of example (2), it is evident that for the three communicators, the referent of I — as well as the referents of the two other possible I’s — was anything but external to the situation; it must have been as clear as the relation between he and Vidor. If the current speaker refers to herself or himself as I and to the fellow communicator(s) as you, it is unlikely that this could ever be a serious problem to co-present communicators. Let us then look at an example from the two-party conversation. (3) HK: I’m beginning to wonder what I do want. Are there any more ads here in the paper? RG: Well, I think they’re only small ads now. But you know, I look at these small ads and I think, well, am I not overqualiªed for them? HK: Yes, but can you ever really be overqualiªed to teach? Example (3) shows, ªrst of all, both speakers referring to themselves as I. Overqualiªed is another case of repetition. With ads we ªnd both substitution and simple
99
100 Sanna-Kaisa Tanskanen
repetition at work: they ªrst substitutes for ads and is then followed by small ads (substitution), which is repeated and ªnally followed by them (substitution). From a collaborative perspective, what is more important than the actual number of cohesive devices is the manner in which surface cohesive devices are utilised in the interaction between the communicators; it is through their usage by the participants in a conversation that the interactional or collaborative capacity of lexical cohesive devices becomes most apparent. Let us therefore consider how reiteration and collocation relations are used by the participants in the conversations. Table 2 shows the distribution of reiteration and collocation pairs into samespeaker and diŸerent-speaker pairs in the conversations. It is clear that the majority of either reiteration or collocation pairs are produced by the same speaker; the total number of pairs is 88 in Conversation 1 and 115 in Conversation 2. The number of pairs produced by diŸerent speakers is 30 in Conversation 1 and 43 in Conversation 2. The diŸerence between the three-party and the two-party conversation is caused by a general increase in the number of pairs in the two-party conversation. The proportions of reiteration and collocation are approximately the same between the two types of conversation, but because there are more reiteration pairs altogether, they are of course mainly responsible for the increase. Table 2. Same-speaker and diŸerent-speaker pairs of reiteration and collocation (per 1000 words) in the conversations Conversation 1 (polyadic)
Conversation 2 (dyadic)
Reiteration Collocation
TOTAL (same speaker)
81 7 88
105 10 115
DiŸerent Reiteration speaker Collocation TOTAL (diŸerent speaker)
28 2 30
38 5 43
118
158
Same speaker
TOTAL
Another diŸerence between the two kinds of conversation is that the collocation pairs in the three-party conversation seem to be almost entirely produced by the same speaker. Only two collocation pairs are produced by diŸerent speakers. In the two-party conversation, the diŸerence between same-speaker collocation and diŸerent-speaker collocation is smaller. What Table 2 appears to be telling us is that in a multi-party conversation speakers tend to rely more on reiteration, especially between speakers, whereas in a two-party conversation collocation is more frequently used, although reiteration clearly dominates there as well.
Cohesion and collaboration 101
Let us look at an example from the perspective of the speakers. (4) HK: No, not at all. I think this is because, partly anyway, because of the heat in Sudan. And, of course there’s no air-conditioning because it’s such a poor country, and it is very di¹cult to work under conditions like that, when the temperature’s — what? 50 degrees centigrade sometimes. RG: It’s as high as that? HK: Yes, very hot indeed. RG: Hm. So what do the children do during that time? Do they do any sort of agricultural work or they just stay at home? HK: I think the majority stay at home. Example (4), from the two-party conversation, shows speaker HK ªrst talking about heat and air-conditioning and temperature (elaborative collocations); temperature is then speciªed as 50 degrees centigrade. Speaker RG joins in with high (speciªcation), and HK corroborates with hot (equivalence). In the following turn, RG uses both stay and home, which are then repeated by HK. Taking into account the fact that collocation relations are subtle, even di¹cult, compared to reiteration relations, the relative scarcity of diŸerent-speaker collocation pairs in the three-party conversation is perhaps not so surprising. It is understandable that conversations with a more demanding communicative setting, i.e. with more speakers, should show a lower number of these relations. A low number of collocation pairs, especially when produced by diŸerent speakers, may therefore re¶ect the demands of the communicative situation. The communicators in the two-party conversation, on the other hand, are not under the same communicative stress, which may explain why more instances of diŸerent-speaker collocations can be found in them. It is more di¹cult to ªnd a satisfactory explanation for the high number of same-speaker reiteration pairs in the two-party conversation; especially the number of simple repetition pairs is strikingly large in these. One explanation might be that the speakers in a two-party conversation would refer to themselves and to their fellow communicator more often than speakers in a three-party conversation. However, the number of ªrst and second-person pronoun pairs is not higher in the two-party conversation compared to the three-party one, so the instances of simple repetition cannot be explained by a diŸerence created by such usage. At this point it can therefore be concluded that the diŸerences between the two kinds of conversation are that the two-party conversation has a higher number of pairs, especially same-speaker simple repetition pairs, and that the three-party conversation shows a lower number of collocation pairs, especially if produced by diŸerent speakers. If a dyadic conversation is regarded as the stereotypical form of
102 Sanna-Kaisa Tanskanen
spoken dialogue, it can be noted that spoken dialogue is characterised by a high number of same-speaker reiterations and that collocation relations are typically produced by the same speaker, but also by diŸerent speakers. If a polyadic conversation, on the other hand, is the starting point, then it can be pointed out that especially pairs produced by diŸerent speakers “favour” reiteration, whereas collocation is rare.
6.
Patterns of cohesion in written dialogue
Table 3 shows the distribution of cohesive pairs in the mailing-list discussion. Table 3. Pairs of reiteration and collocation (per 1000 words) in the mailing-list discussion Mailing list Reiteration Collocation TOTAL
134 17 151
As in the conversations, reiteration relations are more frequent in the e-mail mailing list as well. There are 134 reiteration pairs and 17 collocation pairs, making a total of 151 pairs of cohesion. Comparing the distribution of pairs in the mailing list to that in the conversations, it can be noted that in terms of the overall number of pairs as well as the number of reiteration pairs, the mailing list would be situated between the two-party and the three-party conversations, but closer to the twoparty conversation. Looking at collocation pairs, we can observe that there are more collocation pairs in the mailing list than in either of the conversations (although Conversation 2 is quite close to the mailing list in this respect). At this point it is interesting to consider the distribution of cohesive pairs for the categories of reiteration and collocation. Table 4 presents the number of pairs for each of the categories. As was the case with the conversations, the most frequent relations in the mailing-list discussion are simple repetition relations. Substitution comes second, followed by elaborative collocation, speciªcation and co-speciªcation. Complex repetition, equivalence, generalisation and contrast are each represented by a few pairs. Other collocation relations except elaborative collocation are rare, although the mailing list also shows three pairs of activity-related collocation. Compared to the face-to-face conversations, the distribution of reiteration and collocation pairs in the mailing-list discussion reveals similar tendencies. The heavy
Cohesion and collaboration 103
Table 4. Reiteration and collocation relations (pairs per 1000 words) in the conversations and the mailing-list discussion
Simple repetition Complex repetition Substitution Equivalence Generalisation Speciªcation Co-speciªcation Contrast Ordered set Activity-related collocation Elaborative collocation TOTAL
Conversation 1 (polyadic)
Conversation 2 (dyadic)
Mailing list
71 3 16 – 3 5 11 – – – 9 118
90 1 17 5 2 10 18 – 1 1 13 158
83 5 24 3 3 7 7 2 – 3 14 151
reliance on simple repetition observed in the conversations seems to be a feature of the e-mail messages as well. Let us next take a look at examples from the mailing-list discussion in order to see how reiteration and collocation pairs function there. (5) Strictly speaking, this isn’t about women’s studies but since you are my community, I will ask if anyone can help me to deal with this. A few weeks ago, a young Latino (from the Dominican Republic) who is taking my intro to Latin America history class came to chat in my o¹ce and asked me what I thought of Hitler. I was thrown by the question — I do a lot of work with the Latino/a students, am Latina myself, and Jewish as well. He said that he didn’t understand why Jews thought of themselves as special and didn’t Hitler have a point that they were trying to take over his country. I spent a good hour talking with him about anti-Semitism and genocide, and the things that distinguished the Nazi extermination of the Jews from other forms of oppression in the world. I also told him that it was an issue that aŸected me deeply, that my extended family had lost many people to the Nazi slaughter. Table 4 shows that simple repetition and substitution are the most frequent relations in the mailing list, and example (5) demonstrates why this should be the case. The writer refers to herself several times; hence the repetitions of I. Furthermore, the message centres on the actions of a student, who is mentioned several times (a young Latino – he). There is also a complex repetition pair Jewish – Jews; themselves
104 Sanna-Kaisa Tanskanen
substitutes for Jews, and is repeated by they. Example (5) also shows two equivalence pairs: chat – talking, the Nazi extermination of the Jews – the Nazi slaughter; the ªrst unit of the latter pair is a speciªcation of genocide. Finally, as an example of collocation, asked is related to question by activity-related collocation. It was noted in Section 4 that writers of e-mail messages have a unique feature at their disposal (if their mail system allows it), namely that they can copy text from a previous message into their own message and thus make clear the dialogicality in their message. Studying cohesion in such sequences and comparing the ªndings to those from conversations could reveal if writers of e-mail messages use similar strategies to speakers in a conversation. However, mailing lists apparently vary as regards their tolerance towards text-copying, and the list analysed here does not seem to favour text-copying. This may be due to the fact that the messages tend to be lengthy even by themselves, and thus adding text from earlier messages would produce even longer messages, which many mailing lists tell their participants to avoid. Another factor contributing to the rare use of text-copying on the list may be how long the discussion lasted: the discussion on student bigotry lasted for only a week, with messages being posted at a high speed. Consequently, the list-members may not have found it necessary to remind other list-members of what had been said in previous messages, since they could assume that they would still remember most of the points raised in the discussion. Whatever the reason, text-copying is uncommon in this discussion, for it occurs in only ªve out of the 42 messages analysed. Example (6) illustrates the use of cohesion in one of such messages: (6) > This list has changed a bit since I last logged on, as I didn’t think we > were supposed to have discussions like this…. At the risk of adding further to the list’s exceptionally heavy mail volume, I think I should try to explain why I think this discussion is appropriate for WMST-L. The situation Rosie described is very similar to problems that arise involving oŸensive classroom speech and behavior directed against all or some women. Indeed, a number of respondents have oŸered suggestions based explicitly on their experiences with just such situations. So I do feel this discussion falls within the list’s focus…. Example (6), which is an extract from the beginning of a message, shows a similar reliance on repetition and substitution relations as example (5) above. List and discussion from the copied portion are repeated in the message proper. Both writers also repeatedly refer to themselves as I. In the message proper, situation is repeated, and their substitutes for a number of respondents. Similar tendencies can be observed in the other four instances of text-copying: reiteration relations seem to be favoured, especially between copied portions and
Cohesion and collaboration 105
responses to them. As the message proceeds, collocation relations also appear. On the whole, in the discussion analysed here, the messages which include text-copying do not show diŸerent cohesive relations from messages without text-copying. On the basis of these ªndings, it can be concluded that the mailing-list discussion shows a cohesive proªle quite similar to the conversations. The same relations seem to be frequent in both types of dialogue. In the use of collocation relations, the mailing-list text is closer to the two-party conversation. The number of pairs in the mailing list is not quite as high as in the two-party conversation, but higher than in the three-party conversation. If positioned on a continuum with the conversations, the mailing-list text would thus be situated between the two-party and the threeparty conversation, but closer to the former than the latter.
7.
Conclusion
Let us ªrst return to the diŸerences between the two-party and the three-party conversation. To begin with, the number of cohesive pairs was found to be considerably higher in the two-party conversation. This was true both for reiteration and collocation pairs. When the frequencies of the pairs were examined more closely, it was found that although there was a general increase in the number of pairs, the diŸerence was mainly created by a high number of simple repetition pairs in the two-party conversation. Moreover, a great majority of these pairs were samespeaker pairs. What this tells us about the use of cohesion in spoken dialogue is that cohesion consists mainly of reiteration relations produced by the same speaker. As communicative conditions change and the number of speakers increases, this seems to be re¶ected in the use of cohesive devices: the number of both reiteration and collocation pairs is lower. The overwhelming majority of same-speaker pairs, evident in the two-party conversation, is not repeated in the three-party conversation; instead, the pairs are more evenly distributed across speakers. In the data deemed most collaborative in the sense that it represents stereotypical dialogue, i.e. the two-party conversation, we therefore ªnd a dominance of same-speaker devices. Undoubtedly this is still collaboration: the communicators in a dyadic dialogue can use longer turns in which it is possible to create cohesive relations. The collaboration also manifests itself in the number of collocation pairs: compared to the three-party conversation, there are more collocation relations and they are also produced more often by both of the two co-present communicators. It would be unwise to claim, however, that collaboration is only evident in the two-party conversation. In the three-party conversation, collaboration is present in the interlocutors’ negotiation in order to create a coherent dialogue. For individual
106 Sanna-Kaisa Tanskanen
speakers in a three-party conversation, the demands of the communicative situation mean that their possibilities to produce cohesive relations are smaller, but, on the other hand, it is easier for them to create cohesive relations together with their fellow communicators. It appears that the stereotypical form of dialogue, the dyadic conversation, represents a situation where the use of cohesion is carried furthest. When we remember that the two-party conversation was dominated by cohesion pairs produced by the same speaker, we arrive at the somewhat paradoxical conclusion that the use of cohesion in the stereotypical dialogue is based on monologic strategies. But this is not the whole picture. The speakers in a two-party conversation are responsible for the conversation together, in collaboration, and it seems that the collaboration in a dyadic dialogue enables the speakers to create cohesive relations without any di¹culties. In comparison with dyadic conversation, the higher number of speakers in polyadic conversation apparently changes the communicative situation to such an extent that the cohesive proªle is diŸerent. Looking at written dialogue, i.e. the mailing-list discussion, a slightly diŸerent proªle emerges, which, however, comes very close to the proªle of the dyadic conversation. The mailing list shows a somewhat higher number of collocation pairs, which may indicate both the writtenness of e-mail language and the fact that the production constraints are diŸerent in a a mailing-list situation and a face-to-face situation. Writers of e-mail messages are seldom as pressed for time as interlocutors in a conversation: in the former, communicators can make their contributions to the discussion under less severe temporal constraints than in the latter. In terms of its cohesive proªle, the written dialogue simulates the proªle of the dyadic dialogue, and it is interesting to consider the two from the perspective of collaboration. The contributions of the communicators in the mailing-list discussion are deªnitely the most monologic in the data consisting of dialogues: there is nobody to interrupt a writer’s message (or turn in the discussion), which can even be rewritten and polished before it is sent to the list. The use of cohesion in the mailing-list messages is therefore mainly monologic, which is also the case with the dyadic conversation, where cohesion is dominated by same-speaker pairs. In both types of interaction, similar cohesive strategies are favoured. If cohesive devices can act as signals of collaboration, as has been argued in the present paper, communicators in a dyadic conversation and a mailing-list discussion are collaborating with their fellow communicators using the same strategies. Although there are diŸerences between the two as regards the number of communicators and especially the context (e.g. spoken versus written medium, co-presence versus no co-presence), the communicators’ interaction with their fellow communicators and contexts leads to a similar outcome.
Cohesion and collaboration 107
Notes 1. In the use of the terms text and discourse, I follow de Beaugrande (1997: 10, 21): text refers to a dynamic (spoken or written) communicative event including linguistic, cognitive and social factors (with the focus of the analysis on the ªrst of these), while discourse is the umbrella term for interconnected sets of texts. 2. It can also be argued that it is simplistic to assume that collaboration or dialogicality automatically imply agreement between the communicative parties. Indeed, it may be more important to consider and even attempt to manipulate one’s co-communicators in a situation of disagreement. 3. The examples in Section 3.2 come from texts from various corpora: the London-Lund Corpus (LLC), the British National Corpus (BNC), the Lancaster IBM Spoken English Corpus (SEC) and a corpus of e-mail mailing-list messages collected by the author. In Sections 5 and 6, three of these are analysed in detail. In order to make the examples more readable, prosodic information and other coding have been removed from the texts; the punctuation in the BNC and SEC texts has, however, been retained. 4. Halliday and Hasan also draw attention to the fact that the meaning of lexical items depends on the context. They say that …each occurrence of a lexical item carries with it its own textual history, a particular collocational environment that has been built up in the course of the creation of the text and that will provide the context within which the item will be incarnated on this particular occasion. This environment determines the ‘instantial meaning’, or text meaning, of the item, a meaning which is unique to each speciªc instance (Halliday & Hasan 1976: 289; see also Hoey 1991: 220). 5. Working within the ªeld of cognitive semantics, Allwood (1998) formulates a similar argument about semantics which is “cognitive, dynamic and context-sensitive”. He maintains that Meaning and concepts are primarily taken to be cognitive phenomena and are studied in terms of operations on information rather than as static entities. The operations are context-sensitive, so that meaning is seen as determined by operations which are sensitive to and make use of linguistic and extralinguistic context (Allwood 1998: 1). In Allwood’s model, all linguistic expressions have a “meaning potential”, which includes “all the information the person can associate with the expression” (Allwood 1998: 2; see also Linell 1998: 118–121). Allwood sees the context-free meanings of expressions as “activation potentials”, the activation of which into actual meanings is in¶uenced by the meanings of other related expressions and the extralinguistic context, and guided by cognitive operations. Although Allwood’s model is intended for analysing conceptual structures and therefore has very diŸerent aims if compared to McCarthy’s study or the present analysis, it nonetheless shares a similar starting point, namely the conviction that meaning is made in context.
108 Sanna-Kaisa Tanskanen
6. However, the relation between the units need not be totally haphazard. Jordan (1984; 1992) argues that a trigger, which is usually a repetition of the previous topic (unit), can be used to clarify the association between a unit and its re-entry (related unit). Let us consider the example in a slightly modiªed form: …on the topic of the ªring of Christina JeŸrey as US House of Representatives historian. The reason she was ªred was because she complained in 1986 that a proposed Jr. High curriculum on the Holocaust was not balanced or objective because it did not include the Nazi point of view on the Holocaust… Although the example in its new form may not sound as natural as in its original form — Jordan (1984) observes that there must be a good reason for including the trigger — we can see that it still makes sense even with the repetition of the Holocaust. If there was no relation between the units, the result with the trigger would hardly be so satisfying. Consequently, a “trigger-test” can be helpful in verifying some elaborative relations. 7. The numbers shown in the tables are normalised to a text length of 1000 words. This makes it possible to compare frequencies of cohesive pairs in texts of diŸerent lengths in a reliable way (for details of the normalising procedure, see, for instance, Biber 1995). 8. In the extracts dicussed only those lexical units which are related to other units in the current extract are italicised and discussed, because discussing units related to units outside the extract would be misleading.
References Allwood, Jens 1998 “Semantics as meaning determination with semantic-epistemic operations”. In Cognitive Semantics: Meaning and Cognition, J. Allwood and P. Gärdenfors (eds), 1–17. Amsterdam and Philadelphia: John Benjamins. Anderson, Anne 1995 “Negotiating coherence in dialogue”. In Coherence in Spontaneous Text, M. A. Gernsbacher and T. Givón (eds), 1–40. Amsterdam and Philadelphia: John Benjamins. Anderson, Anne, Robertson, Alasdair, Kilborn, Kerry, Beeke, Suzanne and Elizabeth Dean 1997 “Dialogue despite di¹culties: a study of communication between aphasic and unimpaired speakers”. In Conversation: Cognitive, Communicative and Social Perspectives, T. Givón (ed.), 1–39. Amsterdam and Philadelphia: John Benjamins. Beaugrande, Robert de 1997 New Foundations for a Science of Text and Discourse: Cognition, Communication, and the Freedom of Access to Knowledge and Society. Norwood, NJ: Ablex. Biber, Douglas 1995 Dimensions of Register Variation: A Cross-Linguistic Comparison. Cambridge: Cambridge University Press.
Cohesion and collaboration 109
Brown, Gillian 1995 Speakers, Listeners and Communication: Explorations in Discourse Analysis. Cambridge: Cambridge University Press. Bublitz, Wolfram and Uta Lenk 1999 “Disturbed coherence: ‘ªll me in’”. In Coherence in Spoken and Written Discourse: How to Create It and How to Describe It, W. Bublitz, U. Lenk and E. Ventola (eds), 153–174. Amsterdam and Philadelphia: John Benjamins. Carter, Ronald 1987 Vocabulary: Applied Linguistics Perspectives. London: Allen & Unwin. Clark, Herbert H. 1996 Using Language. Cambridge: Cambridge University Press. Clark, Herbert H. and Deanna Wilkes-Gibbs 1986 “Referring as a collaborative process”. Cognition 22: 1–39. Coates, Jennifer 1995 “The negotiation of coherence in face-to-face interaction: some examples from the extreme bounds”. In Coherence in Spontaneous Text, M. A. Gernsbacher and T. Givón (eds), 41–58. Amsterdam and Philadelphia: John Benjamins. Dickinson, Connie and Talmy Givón 1997 “Memory and conversation: toward an experimental paradigm”. In Conversation: Cognitive, Communicative and Social Perspectives, T. Givón (ed.), 91– 132. Amsterdam and Philadelphia: John Benjamins. DuBartell, Deborah 1995 “Discourse features of computer-mediated communication: ‘spoken-like’ and ‘written-like’”. In Organization in Discourse [Anglicana Turkuensia, 14], B. Wårvik, S-K. Tanskanen and R. Hiltunen (eds), 231–239. Turku: University of Turku. Fox Tree, Jean E. 1999 “Listening in on monologues and dialogues”. Discourse Processes 27(1): 35– 53. Halliday, M. A. K. 1989 Spoken and Written Language. Oxford: Oxford University Press. 1994 An Introduction to Functional Grammar. 2nd ed. London: Edward Arnold. Halliday, M. A. K. and Ruqaiya Hasan 1976 Cohesion in English. London: Longman. Hasan, Ruqaiya 1984 “Coherence and cohesive harmony”. In Understanding Reading Comprehension: Cognition, Language, and the Structure of Prose, J. Flood (ed.), 181– 219. Newark, Delaware: International Reading Association. Hoey, Michael 1991 Patterns of Lexis in Text. Oxford: Oxford University Press. 2001 Textual Interaction: An Introduction to Written Discourse Analysis. London and New York: Routledge. Jordan, Michael P. 1984 Fundamentals of Technical Description. Malabar, Florida: Krieger.
110 Sanna-Kaisa Tanskanen
1992
Linell, Per 1998
“An integrated three-pronged analysis of a fund-raising letter”. In Discourse Description: Diverse Linguistic Analyses of a Fund-raising Text, W. C. Mann and S. A. Thompson (eds), 171–226. Amsterdam and Philadelphia: John Benjamins.
Approaching Dialogue: Talk, Interaction and Contexts in Dialogical Perspectives. Amsterdam and Philadelphia: John Benjamins. Martin, James R. 1992 English Text: System and Structure. Amsterdam and Philadelphia: John Benjamins. McCarthy, Michael 1988 “Some vocabulary patterns in conversation”. In Vocabulary and Language Teaching, R. Carter and M. McCarthy (eds), 181–200. London and New York: Longman. Nyyssönen, Heikki 1992 “Lexis in discourse”. In Nordic Research on Text and Discourse: Nordtext Symposium 1990, A-C. Lindeberg, N. E. Enkvist and K. Wikberg (eds), 73– 80. Åbo: Åbo Akademi Press. Schober, Michael F. and Herbert H. Clark 1989 “Understanding by addressees and overhearers”. Cognitive Psychology 21: 211–232. Sinclair, John 1998 “The lexical item”. In Contrastive Lexical Semantics, E. Weigand (ed.), 1–24. Amsterdam and Philadelphia: John Benjamins. Tanskanen, Sanna-Kaisa 2000 Collaborating towards Coherence: Lexical Cohesion across Spoken and Written Discourse in English. Doctoral Dissertation, Department of English, University of Turku. Traxler, Matthew and Morton Ann Gernsbacher 1995 “Improving coherence in written communication”. In Coherence in Spontaneous Text, M. A. Gernsbacher and T. Givón (eds), 215–237. Amsterdam and Philadelphia: John Benjamins. Weigand, Edda 1998 “Contrastive lexical semantics”. In Contrastive Lexical Semantics, E. Weigand (ed.), 25–44. Amsterdam and Philadelphia: John Benjamins. Wilkes-Gibbs, Deanna 1995 “Coherence in collaboration: some examples from conversation”. In Coherence in Spontaneous Text, M. A. Gernsbacher and T. Givón (eds), 239–267. Amsterdam and Philadelphia: John Benjamins.
Part II
Metadiscourse and discourse markers
Gestural and symbolic uses of the deictic here in academic lectures Julia Bamford University of Rome
1.
Introduction
Academic lectures are still the main vehicle of instruction by which teachers in institutions of higher education impart their discipline to potential members of their discourse community. Despite this widespread use there has until recently been very little research of an empirical kind into what actually goes on in lectures and the linguistic and visual means lecturers use in their teaching. I have argued elsewhere (Bamford, 2000, Bamford forthcoming) that the persistence of the lecture, despite a move towards increasing informality, can be explained by its interactive nature, the main goal of which is to establish contact with the students. The lecture serves also the purpose of starting a conversation with students who are being co-opted into the discourse community, a conversation which starts oŸ highly asymmetrical and gradually becomes a conversation among equals as the process of assimilation into the discourse community takes place. This process is facilitated initially by familiarising the student listener with the concepts, information, methodologies and practices of the discipline. Furthermore a crucial component involves the introduction to the linguistic conventions of the discipline — specialised lexis, rhetorical conventions, recurrent use of typical grammatical structures and collocational patterns to cater to the student’s communicative needs. The economist McCloskey (1985) has discussed both the importance of a conversation between fully ¶edged members of the discourse community of economists and the importance of using the “right” language at the “right” time in order for innovative ideas to be accepted. Linguistic form is recognised as being an important part of a discipline’s genre knowledge (Hyland 2000). Academics from all disciplines go to great lengths to calibre their discourse to the needs of their students, tailoring their talk to make it as understandable as possible. Extract (1) below, from an undergraduate biology lesson, is an example of this.
114
Julia Bamford
(1) Michigan Corpus of Academic Spoken English (MICASE) one kind of cause the cause that people typically think about , when they’re talking about biological causes is what we would call the proximal . the proximal cause just means the close cause. the immediate cause , same- it shares the same root as w- words like approximate it’s close, nearby. so in terms of biopsychology when you’re looking at proximal causes, you’re looking at how current changes in physiology or structure, change behaviour. Extract (1) shows the lecturer’s high degree of sensitivity to the linguistic needs of his students; this is evidenced not only by his explaining a word which has a speciªc meaning in biology, but also by linking it with words that the student might be more familiar with and even providing an etymological, linguistic explanation — “the proximal cause just means the close cause . the immediate cause, same — it shares the same root as w- words like approximate it’s close , nearby”. This type of linguistic sensitivity is widespread in all the lectures analysed in this paper and is an instance of what is known as recipient design in Conversation Analysis (Goodwin 1981). Recipient design occurs because speakers invariably tailor their talk to the interactive needs of their listeners.1 Lecturers display a particular care and attention to student listener needs and the linguistic manifestations of this constitute one of the characterising features of lectures as a genre. Among the linguistic features most frequently used to cater to the listeners’ needs are metadiscourse, question/answer sequences, repetition and reformulation, marked prosody and the features also include the feature which will be the main focus of this paper — the use of deixis.
2.
Corpora and methodology used in this analysis
The data analysed in this paper come from various sources. The principal source is a small corpus of academic lectures in English in various sub-disciplines of economics- economic theory, trade and growth, development economics, rational expectations, employment and macro and micro economics. These lectures (a total of 185, 000 words) were recorded at the University of Siena where guest lecturers are frequently invited to teach modules in undergraduate and graduate courses and where all the graduate programmes are in English. Many of the lecturers were nonnative speakers. Initially the lectures were only audio-recorded but during the course of the construction of the corpus it became obvious that visual codes were being used frequently during the lectures and that the interface between the visual and the verbal was an important feature. Thus videorecording became the standard practice in the corpus construction. All of the lectures have been transcribed using
Gestural and symbolic uses of the deictic here
a modiªed version of the conventions developed in the conversational analysis tradition (Button and Lee 1987) with annotation on visual contextual features relevant to the discourse. The availability of the Michigan Corpus of Academic Spoken English, which by now has a total of nearly 2 million words, has permitted a wider range of academic disciplines to be examined. This corpus is in continual expansion and allows the selection of the type of spoken academic discourse to be analysed. The examples analysed in this paper are all lectures (as opposed to seminars, tutorials, o¹ce hours etc.) so the actual number of words used here is much lower. The lectures have all been audio-recorded at the University of Michigan and include a wide variety of academic disciplines in the arts, sciences, social sciences and humanities. The transcription conventions are slightly diŸerent from those used in the Siena corpus but are su¹ciently intuitive to need no particular explanation. For comparative purposes I have also looked at spoken discourse in a selection from the British National Corpus (the BNC Sampler). The methodology I have used in this analysis is hybrid. I have used both very small and larger corpora but apart from my brief excursus into the British National Corpus for comparative purposes I have used the corpora in rather unorthodox ways. I have collected the Siena corpus and to a certain extent also transcribed it personally. This permits an in-depth analysis of how lecturers use visual materials and a close examination of the relation between deictics and gesture. The Michigan corpus being only audio-recorded did not permit this type of analysis. However, the annotation, which indicates such things as when an overhead is being shown or when the speaker writes on the board or any other type of visual is being used, enables a clear picture of what is going on to emerge. Moreover the size (nearly one million words) and the variation in types of lecture, in addition to the key word in context feature, permitted both comparisons with the Siena corpus and generalizations to be made on a sounder foundation.
3.
Visuals in lectures
In this paper I will be looking at patterns of the deictic here primarily in lectures. Some, though by no means all, of the most interesting patterns of here occur in the discourse surrounding visuals in lectures. By visuals I mean anything presented in the form of graph, diagram, table, mathematical model or symbols, map, painting etc. which a lecturer uses to amplify his powers of explanation in the verbal form. Visuals usually serve to assist the verbal, but they can constitute an essential part of the presentation of concepts. In fact, in one case the entire lecture consisted of mathematical models constantly being written on the board
115
116
Julia Bamford
and, at the same time, discussed. This illustrates the di¹culty of claiming an ancillary role for visuals. A wide variety of visuals are used in lectures. In scientiªc subjects, including economics, they tend to be used to present more abstract concepts which are di¹cult to explain verbally whereas in the arts they tend to have a more practical illustrative function. Another important macro-aspect of visuals is that the language which surrounds them is, generally speaking, more implicit. The language of lectures typically is more explicit than that used, for example, in spontaneous conversation. This explicitness derives not only from the fact that lecturers and their students may not know each other well, so that what is normally presupposed among people who are familiar with each other may have to be made explicit, but also the pedagogic purpose imposes the spelling out of concepts and terminology to render them accessible to the student listener. Example (1) above is a good example of this. The explicitness of lectures is ensured by such features as meta-discourse, repetition and reformulation and the provision of visual aids as is exempliªed in the extract below. (2) Michigan Corpus of Academic Spoken English (MICASE) so i’m gonna start with some deªnitions, and and i don’t really , care to make you memorize these, deªnitions, but, at least if i put them down like this, uh we’ll start from a common place. my deªnition of philosophy is <WRITING TYPE= “ON OVERHEAD”> any thought …directed, toward evaluation …of, sensory experience, or mental constructs. what do i mean by that? basically any thinking you do that’s trying to ªgure out what’s going on and what you’re seeing and hearing and feeling or, what you can think about and conceive, any of that activity i would call, broadly in the realm of philosophy …science …i would deªne as <WRITING TYPE=“ON OVERHEAD”>a branch of philosophy .. seeking to organize , and explain ..material phenomena. > PAUSE DUR=“:07”> okay? so the point of science is to understand, physical things that you can see and feel and measure and touch and hear Extract (2) illustrates how meta-discourse “so I’m gonna start with some deªnitions” and meta-discursive announcements “my deªnition of philosophy is” render the discourse intentions of the lecturer explicit. The repetition of the lexical item “deªnition” ªxes the student’s attention on a salient point by providing redundancy. Furthermore a question such as “what do i mean by that?” followed by an answer in the immediately contiguous slot “basically any thinking you do that’s trying to ªgure out what’s going on and what you’re seeing and hearing and feeling or, what you can think about and conceive” also serves to signal to the listener the need to pay particular attention to the answer (see Bamford 2000). These are all explicit signals to the student that what follows is important information.
Gestural and symbolic uses of the deictic here
The discourse surrounding visuals in lectures however, suddenly becomes more implicit, that is it relies on such things as ellipsis, deixis and pronominal reference so that we ªnd stretches of transcribed lectures in which it may become di¹cult for the analyst to understand what was really going on in the original interaction. It was this aspect that led me to attempt to identify the characteristics of language surrounding visuals in lectures. The extract below (3) demonstrates some of the typical features of discourse in which visuals are being produced and used more or less simultaneously. The importance of the local context to the understanding of the referents is essential. In this case the local context consists of the visual the lecturer draws on the blackboard. Without access to the local context we ªnd it di¹cult to understand, for example, utterances such as “So this is the variable to be explained” or “So we have to put those other things in there” or indeed most of the other talk in this extract. The arrows indicate implicit features such as deixis or reference (3) P. K. Siena 1998 Whatever data you have maybe you take some average and you ªnd a ⇒ growth rate . you remember I told you percentage growth ah now this is uhm the deep end of the (9 secs writes on board) ⇒ So this is the variable to be explained (3secs) This is what we have to explain And you can have many explanatory variables (22 secs [writes on board]) Now what we are trying to explain is really how does trade aŸect growth so we should just take some trade variables and uh these are all you would say some equal to depends upon all these dep uh independent variables How many trade variables we take ah . ªrst of all what does growth depend upon (5 secs) Growth depends on many things other than trade . right Trade is not the only thing that (?) a country grow there are many other things So we have to put those other things in there (17 secs) We can call them control variables (18 secs) And obviously the control variables we would have Capital LABour EduCAtion and we can take the growth of capital growth of labour growth of education whichever way What about the control variables in there (4 secs) And uhm those will be the control variables and there are some trade related variables which we are really interested in ⇒ We’re not really interested in these but we have to put them in into the regression just to take out the fact of education capital etcetera
117
118
Julia Bamford
So whatever will be left here will be truly ah an explanation of how trade aŸects (4 secs) Now the trade kind variables (20 secs writes on board) We can have many (2secs) we could have So we could have many other possibilites ⇒ We call this x or growth and we could have x by y export short or we can even have you know if you really want to explain why countries grow through trade maybe uh their terms of trade (2 secs) Remember terms of trade the price of export (talking while pointing at blackboard) ⇒ Or you could have even growth . the star means the rest of the world OK board eccetera Now what I’m saying is that these are (3secs) a candidate .(2secs) candidate means they’re possible Student: What what does that mean? (incomprehensible 15 secs) P. K. Possible OK? Questions? ⇒ So here you can put in as many as you can think (3 secs) What would be available we can put because what we are doing in empirical analysis we are trying to ªnd is there any relationship or not . OK ⇒ So if you put in something here and you say try this that’s ok I’ll try it so we’re not restricted to just these these are good candidates (oh yes?) In classrooms but also in laboratories (Ochs et al. 1996; Roth 2000) where talk is about graphic representations in the form of visuals it is often the character of gesture as a form of graphic representation that is being exploited. Utterances regarding these visuals often combine with gesture, not only to reinforce the spoken mode, but also to enact iconically what the visual is about. Gesture makes salient certain parts of the visual (“So we have to put those other things in there”) and deixis reinforces this. Gesture and deixis often work together in the discourse surrounding visuals reinforcing each other to make the salient parts of the visual explicit.
4.
Deixis
Deixis can be considered as the prototypical case in which grammar and pragmatics interlock; a grammatical category (a deictic) cannot usually be understood without
Gestural and symbolic uses of the deictic here
pragmatic recourse to contextual factors. All language is understandable in relation to a context but in the case of deixis this interdependence is fundamental and meanings can only be established in close relation to the situation of the deictic item in the utterance and the situation of the utterance in the wider context. The primary archetypal function of deixis is associated with the gesture of pointing in space and as Matras (1998: 394) says, “there appears to be general agreement among pragmatic and functional approaches to deixis as regards its role as a device through which this situation-bound gesticulative action of pointing to a physical object is formalized in grammar”. However while this function of a deictic is relatively straightforward there are a whole range of uses of deictics in which the linguistic pointing to a physical object is not present and on this point opinions diverge. It seems as though speakers use deictics irrespective of whether the referent indicated is present or not, moreover physical pointing can be transferred as a cognitive action to a discourse-based linguistic context. Deictics are closely related to another linguistic phenomenon having an indexical function. Anaphora refers to expressions whose meaning can only be understood by reference to how an utterance is situated in a text. Deixis and anaphora are said to diŸer because they are situated diŸerently in discourse and point to diŸerent types of context. Deixis anchors an utterance to a referent which has usually been deªned as external to talk (SchiŸrin 1990, calls this context) while anaphors refer to part of previous discourse internal to talk i.e. a linguistic world or a ‘text’. This distinction as well as that between deixis used when the referent is physically present, and that used when there is a cognitive transferral to a linguistic context, brings up a series of problems. As we will see, when examining real discourse examples, it is not always easy to determine whether we are dealing with deixis in its canonical, physical, pointing form or whether we are in the presence of discourse-based deixis. The distinction between the two can be fuzzy. SchiŸrin (1990) also ªnds that the diŸerence between deixis and anaphora can be blurred because, for example, in some cases a deictic item can also be used anaphorically as in example (4) where ‘these’ is both a deictic and refers anaphorically back to “terms of trade the price of exports” (see longer Extract 3 in the previous section). (4) P. K. Siena 1998 Now what I’m saying is that these are (3secs) a candidate .(2secs) candidate means they’re possible The deªnition of deixis by Lyons manages to encompass both aspects of deixis i.e. the linguistic indication of something that is physically present and also what Fillmore (1997) has called the symbolic use of deixis, in addition to its relation to a spatiotemporal dimension.
119
120 Julia Bamford
By deixis is meant the location and identiªcation of persons, objects, events, processes and activities being talked about, or referred to, in relation to the spatiotemporal context created and sustained by the act of utterance and the participation in it, typically, of a single speaker and at least one addressee. (Lyons 1977 : 637)
However, even the spatiotemporal distinction, seemingly so clear-cut in deªnition, can be problematic when real speech events are examined. This can be illustrated by the two following extracts, (5) and (6), which are so typical of metadiscursive “teacher talk” at all levels of the educational spectrum. In the extracts from economics lectures, here, which is normally thought of as a spatial deictic, is used temporally and can be glossed as ‘at this point in the lecture’ or even ‘at this point in time’. This typical pattern used by lecturers has also been noted by Quirk et al. (1985: 1453). In their discussion of the use of here as an adverbial of time the example they quote shows remarkable similarity to those below. The relation between space and time often involves the use of metaphor based on the notion of movement. We can speak of the future as being ahead of us and the past behind us as though the world were moving. In the examples below we can perhaps see the metaphor of the lecture as an intellectual journey operating with stages, stops for rest or questions. (5) S. B. Siena 1998 I think I will stop here and we will continue our discussion tomorrow (6) Pari K. Siena 1998 Any questions here ªrst Fillmore (1997) distinguishes between various types of deixis: – person deixis which anchors the identity of interactants in a communicative situation – place deixis which can be interpreted by knowing the place or places in which the interactants are located – time deixis which is anchored to the time at which the communicative act took place, and further subdivided into encoding time (the time at which the message was sent) or decoding time (the time at which the message was received) – discourse deixis is the matrix of language in which the utterance takes place: that is the discourse surrounding the deictic item – social deixis as characterised by those terms of address which typically mark social intimacy and distance, honoriªcs, politeness or insults Deixis, especially spatial and temporal deixis, can also be classiªed as proximal or distal and as such represents prototypical examples of how language encodes
Gestural and symbolic uses of the deictic here
degrees of spatial context. Here and there are examples of respectively proximal and distal deixis. Although these serve to indicate relative nearness or distance of the referent with respect to the speaker they can also be used to indicate other more complex attitudes. Glover (2000), for example, following Hanks (1992) ªnds that proximal reference may be used to introduce a topic into the discourse or negotiate it while distal reference is often used with established topics. In these cases it is not so much the actual spatial distance that is being encoded in the deictic but rather the speaker’s attitude.2 From this brief discussion of deixis we can see several problematic issues emerging; namely the distinction between deixis and anaphora and the distinction between when the deictic item refers to a physically present entity and when it refers to something more abstract or at least only present cognitively in the current discourse. This discussion of the role of here in academic lectures hopes to contribute to clarifying these issues.
5.
Symbolic, gestural and anaphoric deixis
Fillmore (1997) in his discussion of deixis makes the useful distinction (also taken up by Levinson1983) between what he calls the gestural, symbolic and anaphoric uses of a deictic.3 This is particularly applicable to the spatial adverbial here. The gestural use of a deictic expression can be considered the unmarked one, especially if we consider the various deªnitions of deixis discussed above (see Section 3) which consider the deictic as anchored to a social context and in some way as pointing to some aspect of the space and time in which the communicative act takes place. It can be deªned as being interpretable by a participant who is monitoring the physical aspect of the communicative situation. This is borne out by the already mentioned di¹culty for the analyst who is not physically present during the interaction, nor has access to video-recorded material, to decide the referent of deictic expressions such as here . It is by no means always obvious from a corpus or even from one’s own audio-recorded materials what here or any other deictic refers to. Often other linguistic or prosodic cues help interpretation because, as in so many communicative situations, the participants use a wide range of codes (verbal, prosodic and gestural) to communicate. Moreover, even these single channels can be reduplicated to ensure that the message is got across by such devices as repetition and reformulation, the use of changes in both rhythm and pitch or the use of both eye and body movement. Gestural use of a deictic anchors the utterance to the physical context most closely in that the utterance can only be fully interpreted by reference to contextual features. In the case of here these features are spatial, relating the speaker, the
121
122 Julia Bamford
listeners and the point in space referred to. Deixis is organized in general in an egocentric way so that the anchorage point is the speaker and the point referred to as here is to be interpreted in relation to him. In the video-recorded lecture in economics (Extract 7) there is a coincidence in the deictic here of several codes operating at the same time — the verbal, the gestural and the prosodic. This practically simultaneous use of three codes takes place because the lecturer refers to the visual he has previously drawn on the board, by using the deictic here to draw attention to a particular point in it, but he also marks this feature of his talk as especially salient by attaching to the deictic both gesture and marked prosody. The lecturer traces a spatial relation between himself, the point in the diagram on the board and the listener. He has previously created a local context (the visual on the board) which he refers to deictically. (7) Siena 2000 We could deªne a little circuit in here Where diŸerent ªrms in the sector have preferences for each other’s goods The symbolic use of a deictic expression on the other hand, enables a participant in an interaction to interpret this, using his knowledge of certain aspects of the communicative situation, whether this knowledge derives from current perception or not. In other words, in this case the pointing does not, and in many cases can not, occur physically but the deictic is interpreted cognitively. Identifying symbolic deixis or rather distinguishing it from gestural deixis can present problems when we come to examine occurrences from corpora. In some cases the demarcation line between a gestural and a symbolic use is fuzzy, but since fuzzy boundaries are characteristic of many linguistic phenomena it should not constitute a valid reason for abandoning the distinction. Levinson (1983: 79) has deªned the symbolic use of deixis as “the pragmatically given unit of space that includes the location of the speaker at CT (coding time)” but this could be extended to include the encyclopaedic knowledge of the participants and also the cognitive space. In other words, what distinguishes symbolic deixis from gestural deixis is that the former can have a referent that is only present cognitively, but knowledge of this referent must be shared by the participants. Extract (8) illustrates the symbolic usage of here where what is intended by here is somewhat fuzzy. In fact the referent of here in this case is an indeterminate ‘what happens at this point in the lecture’ or ‘what I am trying to achieve in this lecture’. (8) Michigan Corpus of Academic Spoken English Alright this is just a sketch i’m not doing this in detail, i’m just trying to give you a ¶avor for what happens here
Gestural and symbolic uses of the deictic here 123
Although there is often a certain amount of fuzziness, in fact there are many cases in which the symbolic use of here can be glossed as ‘in this place’. Its referent is always more vague than the gestural use and typically it is not made salient by special prosody or gesture. This semantic fuzziness of here can lead to such uses as illustrated in extract (9). (9) BNC Sampler (utterances) And here’s me with one eye In extract (9) the notion of space or place has got lost and in its place we ªnd a rather formulaic utterance (it can be transformed into here’s x substituting ‘me’ with a proper name) which speakers use to make narrative sections of their discourse lively and give it immediacy. The present tense of the verb to refer to past time also reinforces this eŸect. This semantic fuzziness can be seen in the lecture corpus also, as illustrated in extract (10). (10) P. K. Siena 1998 So I won’t go into good econometrics or bad econometrics out here but you have to have a trade to make that your regressions are good . or at least not obviously bad. The referent of “out here” in extract (10) is very vague and does not refer to a physical space but rather refers symbolically to an intellectual space. It can be glossed as “in this lecture” (see also the discussion in Section 7). One of the diŸerences between a symbolic and a gestural deictic relates to its relation to context. A symbolic deictic relates to a global context whereas a gestural deictic relates to a more local context. Fillmore (1997) also speaks of the way in which symbolic and gestural uses of deictics can in¶uence their grammatical acceptability. Gestural uses of here are much more ¶exible in their acceptability: for example “take them here” which can be made salient by an accompanying gesture would be acceptable whereas a symbolic use of this type would not. The anchorage of a symbolic use of deixis to the context is much more tenuous, culture-speciªc and requires more processing, while the gestural use of deixis is very closely linked to the immediate physical context and is totally reliant on it for interpretation. A proximal deictic such as here also typically diŸers from a distal deictic such as there in that they are more ªgure-like (Hanks 1993); they are more attached to the speaker than the referent (ªgure vs ground).The idea of ªgure/ground (Hanks 1990) organizes relative distance in visual perception in the ªrst place but it is not limited to this. As I discussed above, the original visual/spatial perception can be extended metaphorically to include temporal and cognitive perception. One advantage of the analogy ªgure/ground is that it focuses our attention on the fact that
124 Julia Bamford
deixis is a framework for organizing the actor’s access to the context of speech at the moment of utterance. Deictic reference organizes the ªeld of interaction into a foreground against a background as ªgure and ground organize the visual ªeld. Anaphoric deixis serves a more textually cohesive function than gestural and symbolic deixis (Halliday and Hasan 1976). The anaphoric use of deixis involves an expression which can only be correctly interpreted by knowing which stretch of the previous discourse it is co-referential with. In other words it refers back to other parts of previous talk and therefore the context which enables its interpretation is a textual one rather than a physically present contextual feature or a shared cognitive context. Among the deictics, here is not used as frequently in an anaphoric function as other deictics like ‘this’ ‘that’ or ‘these’ which frequently refer back to previous discourse. Extract (11) shows a typical usage of the demonstrative deictic (‘this’) which is used anaphorically to refer to the immediately preceding discourse — “the multiplier”, while here is gestural and indicates a particular point in the visual on the board. (11) Siena 2000 EN And the relationship shown here is the multiplier So this is precisely the simple theory of eŸective demand (2secs) There are, however, cases in which the deictic here is used anaphorically in the corpus, as in the following example. (12) Siena 2000 EM So when we’re asking . is the path unique . here is a place where you have to make an extra assumption The referent of here in this extract is the preceding text “is the path unique”. The anaphoric use of here ensures cohesion between the various parts of the text.
6.
Gestural uses of the deictic here in lectures
Gestural uses of deictics seem to have certain characteristics which diŸerentiate them from symbolic uses. They are anchored both to the speaker and to the interactional and physical space in which he is situated, also called by Hanks (1992) the deictic space. The context to which the deictic here used gesturally is anchored, is local and the referent is most typically concrete rather than an abstraction, an idea, an emotion, a feeling or part of the common cognitive baggage of both speaker and listener. Because gestural uses of deixis are speaker-centred and closely context-anchored, they are typical of work situations and interactions where some
Gestural and symbolic uses of the deictic here 125
physical activity is taking place along with the talk. In these interactions the spatial relation between the speaker and the referent is often one of close physical proximity involving action. In the lectures examined in this paper the activity is typically writing on the board or overheads, the creation of visual materials in which spatial relations play an important role. Thus here can refer, for example, to a point in a graph, table, diagram, map, painting, picture, photograph or symbolic representation. This is the archetypal use of here as a verbal pointer, indeed it is often accompanied by a pointing gesture and can also be prosodically marked. There is a growing interest in the empirical study of deixis in real interactional situations (Hanks 1990, 1992; Goodwin 1996, 2000; Hindmarsh and Heath 2000; Glover 2000). All of these studies look at real interactional situations where, along with verbal interaction, in many cases work activities or other physical activities are taking place at the same time. Many of these actions also involve interaction about visually perceptible aspects of the context, such as monitors or hopscotch tracks drawn on the pavement. Hindmarsh and Heath, for example, examine a complex technological work environment in which participants in their work routines have to make reference to and mutually negotiate what constantly changing objects on screens and documents actually mean. They raise the problem of how participants identify the referent and criticize standard deªnitions of deictics as pointers, because these fail to explain how listeners identify the object, indicated by a deictic, out of the abundance of objects in the local context. They argue that bodily actions (gestures) are a resource often taken up by the interlocutors to make clear the referent of a deictic. Lyons (1977: 575) has called gesture deixis at its purest and other writers on the subject emphasize this point (Levinson 1983: 65). Yet until recently very little empirical research has looked at the interface between the verbal and the gestural in the speaker’s attempt to encourage or direct the other participants to look at a particular feature of the physical context. In the corpus of video-recorded lectures there is a high frequency of occurrences of the deictic here which is linked to a gesture. Furthermore here occurs in both the Siena and the Michigan corpora in its gestural form more frequently than in its symbolic form; it is used to indicate a spatially proximal object (or feature of it) present in the local context. Typically this consists of talk surrounding a visual as in the example (13). (13) F. P. Siena 2000 So. The consumption goods . then you have . if I understand correctly . you have a period . and the consumption goods come out again here In extract (13) from an economics lecture, the speaker has drawn a diagram on the board which represents a complex economic question graphically. He models an economy at various points in time and refers to a point in this model where
126 Julia Bamford
consumption goods appear. His referent is a particular point in the local context which he has created by drawing it on the board and this point is indicated deictically by the word here. The point referred to is physically present and can only be understood by the persons actually taking part in the interaction (or by watching the video of the lecture). This is therefore clearly a gestural use of the deictic. As Hindmarsh and Heath (2000) have also noticed as regards a working environment, in many of the video-taped lectures in the corpus there is a simultaneous use of verbal, gestural and prosodic codes to indicate the referent. Lectures are a peculiar type of interaction where almost invariably one speaker holds the ¶oor throughout, but to compensate for this privilege he must organize the talk with maximal recipient design. The frequent intermingling of the three semiotic codes testiªes to this, as does the use of visual means to clarify and explain verbal information. Gestural and prosodic features often accompany talk, in particular when deictics like here are used to refer to a point in visual space. (14) Siena 2000 EN And . one of the . interesting thing about . the initial diagram (bells ring) is that this is a diagram of a two . sector eŸective demand model And the relationship shown here is the multiplier So this is precisely the simple theory of eŸective demand (2 secs) In segment (14) taken from a lecture on economic theory, the speaker accompanies his utterance “the relationship shown here” with a gesture by which he indicates with his outstretched arm and index ªnger, the point in the mathematical model on the board where the demand relationship occurs. Mathematical models and diagrams are important analytical tools in economics at all levels, from ªrst year undergraduate to research seminars. In particular in lectures they present a visual element which helps to render abstract notions more accessible. However, the student listener has to be guided in his reading of these models and one way a lecturer can do this is by indicating both verbally and gesturally the salient points. In the above example the salient point (the demand relationship) is indicated both verbally by here and gesturally. The body movement co-occurs with the talk and the listener sees the two as semantically related. He uses both to help interpret the deictic here. In extract (15) the lecturer provides a triple indication of the referent: the deictic here, a gesture indicating the salient point in the diagram and the prosodic salience marked by a change in pitch. (15) Siena 2000 EN So . the condition which we looked at in the diagram Is derived here from the two-sector model On the assumption of the golden rule
Gestural and symbolic uses of the deictic here 127
These are provided almost simultaneously, but the body movement required to indicate the point in the diagram is slightly anticipated with respect to the deictic, with its attached marked prosody. The lecturer already begins to move his arm and outstretched hand at the beginning of the utterance “Is derived”. This anticipatory use of gesture signals to the student that he should expect a verbal counterpart and therefore pay attention to the immediately subsequent discourse. (16) P. K. Siena 1998 So if you put in something here and you say try this that’s OK I’ll try it so we’re not restricted to just these . these are good candidates Although we would expect to ªnd gestures working in interactions, such as lectures, to support talk and thus to occur either simultaneously or even slightly after the deictic to which they are attached, in a number of cases we ªnd the gesture preceding the talk. Hindmarsh and Heath (2000) also ªnd similar occurrences in their data and suggest that in these cases talk works re¶exively on behalf of gesture. This is illustrated in extract (16) where the lecturer already has the chalk in his hand pointing to a place in his regression before he starts to say “so if you put in something here and you say try this”. The talk is subsequent to the gesture and rather than gesture being parasitic on talk it seems that gesture plays an important role in the communication, indeed anticipating the deictic item. However gesture and verbal deixis are complementary, at least in the examples in the corpus, although there are occasions in which they can be contradictory (see Lyons1977: 63), as for example when a speaker agrees verbally to a proposition but gestures dissent jokingly to other interlocutors, unseen by the participant who uttered the original proposition. The only example of gestural deixis from the Michigan corpus which I present is extracted from a lecture on the impact of the media on communication. The lecturer explicitly states that she is presenting an overhead with a pyramid divided into segments. This represents the distribution of wealth in the United States and she refers continually to the ªgures and the diagram of the pyramid on the overhead, used to illustrate which percentage of the population receive which incomes. The repeated use of here (in bold in the transciption) mirrors her progress from one segment of the pyramid to another which symbolically represents a segment of the US population. This is a clearly gestural use of the deictic to indicate a point in the symbolic space the lecturer has created. (17) Michigan Corpus of Academic Spoken English that’s about seven, point-three percent of the population. okay? [SU-M: in America? ] and this is the United States, yeah. the United States. alright, now
who’s in here?… between seventy thousand, hundred thousand. who’d we have in here? oh, let’s not forget, my dears, anchormen and
128 Julia Bamford
-women… like, um Diane Sawyer who makes a million bucks a year, mkay? okay, um, alright, then, so we have this very rareªed area where the very rich are, okay? who_ what would be next down here?… we also will have some, obviously we have some lawyers up in here, we’d have some surgeons up in here, (mkay)? how ‘bout in here? seventy to hundred thousand?… some journalists, some college professors, some business people some middle management folks, right? some salespeople. [SU-M: college professors ] college professors, accountants some journalists okay. the group in here, this is about eleven percent. mkay? el- about eleven percent makes this (up). mkay then, and, you know my drawing as you guys know is really bad so this pyramid does not really approximate any of the percentages but you get the idea, okay? alright now we go from fortyªve thousand, to seventy thousand. okay who do we have in here? other college professors, other journalists, okay? um, small business people, tehi- t- high school teachers
7.
Symbolic uses of the deictic here in lectures
Symbolic uses of deictics in lectures typically are related to a global context which is tendentially abstract. The referent of here for example can usually not be indicated physically and therefore both gesture and prosodic salience attached to the deictic are much less frequent. Symbolic deixis tends to be anchored to a shared cognitive context and therefore requires greater processing than gestural deixis. With regard to the lectures in both corpora, the Siena and the Michigan one, symbolic uses of here are understandable without the aid of video tapes, since what is referred to is often not present and visible in the physical context. Symbolic deixis is also tied to metaphorical uses, as in idiomatic expressions such as “she’s not all here” which can be glossed as “she’s rather stupid”.4 In idiomatic expressions the space referred to by here is typically vague and part of the common culture and cognitive space as Fillmore’s deªnition suggests. Deixis is the name given to those formal properties of utterances which are determined by and which are interpreted by knowing certain aspects of the communicative act in which the utterances in question can play a role. (Fillmore 1997: 61)
This deªnition accommodates symbolic deixis as well as gestural while many deªnitions of deixis emphasize the purely pointing role of a deictic and fail to take into account the symbolic function.
Gestural and symbolic uses of the deictic here 129
Not only do idiomatic expressions have a vague referent as regards the deictic here; the symbolic deictics which occur in both the Michigan and the Siena corpora have also a consistently vague referent, as illustrated in example (18). Does the ‘out here’ refer to ‘in this lecture’ or does it refer to a particular point in the lecture? Vague language, as Channell (1994) has argued, is by no means uncommon in academic discourse. In this particular example the deictic expression ‘out here’, is used colloquially and, as such, is another example of the speaker’s attempt to be conversational. In any case it is clearly symbolic since no physical entity exists in the context which could be the referent of here. (18) P. K. Siena 1998 So I won’t go into good econometrics or bad econometrics ah out here but you have to have a trade to make that your regressions are good . or at least not obviously bad The following extracts from both the Siena and the Michigan corpora illustrate a common pattern of the symbolic deictic here where the referent could be glossed as ‘this point in our common cognitive space’. However in all these examples neither a precise meaning nor a precise referent is available.5 Channell’s discussion of vague language is very apt, in part because she argues that it forms a more considerable part of language use than was once supposed and is also present in scientiªc discourse. What is particularly to the point in her analysis, as far as here as a symbolic deictic is concerned, is that the meanings of vague expressions are themselves vague and it is not possible to ªnd a precise meaning for the sake of a tidy analysis. Any meaning they have can be ascribed by contextual factors and listeners need to bring to bear not just knowledge of the lexis and grammar of English but also pragmatic knowledge about how language is used and how it relates to its setting. This vagueness seems to be the characterising feature of symbolic deixis whereas the referent of gestural deixis is always relatively precise. Providing that su¹cient contextual information is available the referent can be pinned down to something deªnite. (19) Michigan Corpus of Academic Spoken English The k-k-key question here is, which eŸect , is related (20) Michigan Corpus of Academic Spoken English He what is he arguing here? What is NATHAN SAYING HERE? (21) EM Siena 2000 The point is the that p c .as deªned here . is insu¹cient to buy the capital goods . which are going to produce this
130 Julia Bamford
(22) EN Siena 2000 Your advanced money . is the sum of money advanced . by implication here . pretty obviously if you calculate money and then compare it to income . is LESS . than . the total gross national product In all of the above examples of symbolic deixis the referent is a point in the cognitive space of the speaker which he wants to share with the listener. However a textual relation is also present, as, for example, in extract (21) where ‘as deªned here’ refers back to a previously discussed deªnition of the mathematical symbols p and c. This linkage is both to the previous discourse and to its argumentative structure. SchiŸrin (1990) has argued that deictics play an important textual role and that anaphora is dependent on deixis. This is because both texts and contexts constitute one another. In other words, language is not ‘situated in’ social interaction (for example Levinson, 1983: 54); rather, language is ‘part of’ social interaction such that language and social interaction both create, and are created from, each other. (SchiŸrin, 1990: 265)
However here can also act cataphorically and project the listener’s attention towards the subsequent discourse. This is another example of the ‘listener-friendly’ recipient design of lectures because the speaker by announcing his plan of action tries to focus the listener’s attention on the discourse that follows. Two examples of this can be found in extracts (19) and (22), where, in the former the speaker announces a question and in the latter here is slotted in a question used to focus the attention of the listener. The following extract (23), from a physics lecture, contains three occurrences of here: the ªrst illustrates the problem that can exist in deciding whether an instance is gestural or symbolic in the absence of visual contextual cues. Does here refer to an experiment which is physically present during the current lecture or does it refer back to an experiment discussed during a preceding one? The second occurrence, on the other hand, is clearly symbolic because it does not refer to something in the physical context but to an idea currently under examination. The phrase ‘the idea here’ can be considered as pre-patterned speech; as I mentioned above, here seems to occur frequently in formulaic language. The third occurrence is also symbolic but is more textual in function since it contrasts the lecturer’s assumption with what happens under CW spectroscopy. Contrastive relations are textually cohesive, linking two parts of an utterance so that they are compared. In this utterance, here serves as a signal of contrast. (23) Michigan Corpus of Academic Spoken English what we had last time on the Ramsey fringes . the physical experiment here and developed by Ramsey, is one in which you have something an atomic
Gestural and symbolic uses of the deictic here
beam, that passes through two ªeld regions. and the idea here is to get a line shape eventually , that will, be narrower in frequency than , you would’ve, had if you did, C_W spectroscopy, and you would’ve been limited by the transit time the amount of time the atom stays in the, ªeld interaction regime. here we’re assuming that the atom really, doesn’t spend very much time in the ªeld Lecturers, as I have mentioned, attend to the recipient design of their talk to make it as accessible as possible to their audience. One of the ways they do this is to make their lectures lively by resorting to linguistic devices which make their listeners pay attention. In particular, the lecturer during the ªrst lesson tries to ‘get to know the class’ and involve the students in a disciplinary conversation. Deictics can be used as involvement strategies to make the audience feel they are participating in the interaction. The lecturer in the introductory biology class below (Extract 24) tries very hard to engage his audience, to create some sort of “class consciousness” and the hub of this engagement is here, that is belonging to this class, forming part of a group. In the discourse prior to this, the lecturer has introduced himself giving a few biographical details, and talked about time-tables, assessment and various other course management matters. In this part of the discourse he wants to create a feeling of group membership among the students attending the class, and this is signalled by the deictic here, where here is not a visible object but a feeling of belonging. (24) Michigan Corpus of Academic Spoken English you guys are a pretty good group a pretty competitive lot, uh one of the perks of teaching at a good school like this is that i get gratiªcation out of talking to high-level students uh like you. somewhere between ªfty_ half and two thirds of you, uh wanna be pre-med pre-dent pre-vet whatever, uh another one ªfth or one quarter of you, are some sort of biology uh molecular organismal ecology School of Natural Resource, uh, ten or ªfteen percent of you are engineers and about ªve percent of you are lost. um, you don’t fall into any one of the above categories. the lost people are are often the most interesting people art majors, um, uh or just uh education majors or people just interested in biology so um if you have taken a diŸerent path or a a odd path to get here, uh recognize that and and recognize that you’re in a competitive class, but um you’re also appreciated and, i like hearing from all of you especially the people that have, uh weird reasons for being here. uh what have we gotten ourselves into? um, i hope and this is one of the advantages of of a relatively smaller group like this is i hope that lecture can be at least a modest amount of back and forth.
131
132 Julia Bamford
A similar sense of involvement and of creating rapport with the audience is present also in extract (25), from the beginning of a lecture from the Siena corpus where a guest lecturer is speaking to graduate economics students with some junior faculty present. Although the referent of here in, ‘It’s a great pleasure to be here’ is usually vague, the cataphoric expansion shows that it refers not only to a particular physical space, but is also associated with a person and various pleasurable experiences which the speaker knows at least some of his audience can empathize with. His reference to here is not just a reference to a room, but to a range of experiences associated with the person the room is named after. (25) Siena EN 2000 Thanks . uh (2 secs) It’s a great pleasure to be here . in a room dedicated to Richard Goodwin uh whom I loved very much and who I loved to visit in Siena . uh . he was a wonderful man and uh . loved . a good bottle of wine and uh . conversation
8. Relative frequencies of symbolic and gestural deixis in lectures and conversation One of the most striking aspects to emerge from an analysis of here in lectures is the disproportion between its gestural uses and symbolic uses. Gestural uses tend to be more common than symbolic uses and to occur in this function in language surrounding visuals. There are a few instances when the lecturer refers to other types of objects such as a window or the projector or to persons present at the interaction by the gestural here but the vast majority refer to visuals. Although it would seem more likely that gestural uses of the deictic here are more common in lectures in scientiªc disciplines than in the arts in general, this is not supported by the evidence from the Michigan corpus. A close reading of the transcripts reveals in fact that many disciplines use visuals, not just those with a mathematical bias. These visuals can be photographs or slides or even extracts from texts in the form of an overhead transparency. However whatever form they take they constitute a local context which is frequently referred to deictically as here. The total number of occurrences of here in lectures in the Michigan corpus is 480 while in the Siena corpus it is 314. As I have mentioned when discussing gestural deixis, visual access to physical context is important for interpreting the referents of here correctly. The Siena corpus was video-taped precisely to provide access to visual context. However the Michigan corpus which consists of transcriptions of audio-recordings is annotated scrupulously so that anything in the context which aŸects the talk is noted down (a good example of this is to be found in
Gestural and symbolic uses of the deictic here
example 2 Section 3). This has meant that in most cases I was able to distinguish between a gestural and a symbolic use of here. Nevertheless the ªgures I present to illustrate the relative proportions of gestural and symbolic uses of the deictic here in lectures must be taken as approximations although they give us an idea of the relative importance of the types of deictic. In the Michigan corpus approximately 70% of the occurrences of here were gestural while in the Siena corpus over 80% fell in this category. This diŸerence can be attributed to the fact that visuals appeared in all the Siena lectures but not in all the Michigan lectures. Tables 1 and 2 below show the frequencies of occurrence of here in individual lectures in the two corpora. A quick glance at the ªgures shows that there is no evidence that scientiªc disciplines are any diŸerent from the arts or the humanities with respect to the occurrences of here. As I mentioned above, this is due to the fact that many disciplines use visual materials in their teaching activities and here clusters around these visuals. The lectures in the Siena corpus all have high rates of occurrence of here and all used visuals throughout the lecture. These were constructed on the board during the lecture and simultaneously commented on. This type of activity gave rise to frequent instances of here. Table 1. Frequencies of occurrence of here in lectures in the Michigan corpus Lecture type Introductory anthropology Medical Anthropology Japanese Literature Practical Botany Intro Biology Intro Communication Media Impact in Communication Macroeconomics graduate Intro Engineering Literature and Social Change Intro Oceanography Intro Psychology Drugs of Abuse Perspectives on the Holocaust Principles in Sociology Statistics Total
occurrences 11 34 21 26 14 41 29 33 45 46 41 13 48 12 29 37 480
no. of words 11549 11864 8567 5994 6907 9696 9684 8373 6174 10069 8527 7744 11049 9172 12219 6762 144350
133
134 Julia Bamford
Table 2. Frequency of occurrences of here in the Siena corpus Lecture type Macroeconomics Development Economics Trade and Development Rational Expectations Game Theory Microeconomics Monetary Economics Trade and Growth Employment Total
occurrences 24 37 31 20 46 43 26 51 36 314
no. of words 12799 12503 18350 19132 20123 15879 24840 9854 10982 133787
The two polar extremes in the distribution of occurrences of here were found in two lectures from the Michigan corpus. A public lecture in physics by a Nobel prize winner (not included here because it was not for students) had 140 instances of here while an anthropology lecture had only 11. Taking into account the slightly higher number of words in the physics lecture we still get approximately one occurrence of here every 100 words while the anthropology lecture has one occurrence every 1050 words. This shows how the use of visuals can raise the frequencies of deictics in lectures. Although the literature discussed in Section 6 shows that deictics are used frequently in workplace interaction and where some joint activity is taking place, like playing hopscotch, what happens to here in spontaneous conversation has not been investigated. In order to compare what happens to here in ordinary conversation with the lectures, I looked at the BNC Sampler and extracted 500 examples of here in utterances from what, as far as I could tell, were spontaneous conversations. In order to ªnd as much contextual material as possible I looked at the surrounding discourse and the description of the speakers and the speech event. This method obviously has several ¶aws but I was interested in getting a rough, unsophisticated measure of the relative frequencies of symbolic and gestural use of the deictic here in conversation. Typical of what I found were the following examples. (26) BNC Sampler I’m gonna put Grant’s tablet up here. Annabel, would you like some milk (27) BNC Sampler Great Chris, come on . Daddy’s got two Here you are then . This’ll do (28) BNC Sampler Good boy .Mm. Open wide, there. Yeah. Here you are then. Yes, very good (29) BNC Sampler Just cos I wanna ªnish something. Here I turn now do I? Yeah
Gestural and symbolic uses of the deictic here
(30) BNC Sampler Yes I know it’s been very very heavy here too, there were certain areas over (31) BNC Sampler Yeah, I’ll give you this dog. Well look here Oh thank you Thank you Michael Example (27) is clearly gestural while the others are all symbolic. However, what distinguishes these examples from the lecture corpora is not so much the fact that there are slightly more symbolic deictic ‘heres’ with respect to gestural ones, it is the type of symbolic here that is very diŸerent to those we ªnd in lectures. Gestural uses of here on the other hand are the same in both lectures and conversations. In conversation many of the symbolic uses seem to form pre-patterned utterances such as ‘look here’ used to attract attention or ‘here you are’ used when giving someone something. Another common pattern is ‘he/she/proper name was here’ where the exact referent is unclear. This lack of clarity does not derive only from the fact that contextual information is missing but also from the fact the referent is purposely vague. This is also true in lectures but this vagueness often relates to the ideational realm while in conversations the vagueness is interactive and social.
9.
Conclusions
Lectures like other professional discourse or talk at work use deictics such as here to indicate points in the common spatial context of both speakers and listeners. Gestural here relies closely on the visible physical context for its disambiguation. Gestural deixis in lectures is almost invariably associated with the use of visuals. This is particularly marked in the corpus of economics lectures recorded in Siena but is also evident in the Michigan corpus. Both gesture and prosody are often to be found associated with deictics although they vary in the closeness of their attachment to these. I have discussed in fact, in Section 6, examples of gesture which precede the deictic and I agree with Hindmarsh and Heath (2000) that they do not always play a merely ancillary role in the economy of the interaction. Gestural deixis in the corpora has a precise referent which is interpretable when the visual context is available. The referents of symbolic deixis on the other hand are less precise since they belong to the common cognitive space of both the speakers and their student audiences. This does not necessarily mean that listeners have di¹culty in interpreting them but if asked to deªne what they understood by symbolic here they might have di¹culty in pinning down an interpretation exactly. The referent of symbolic here in lectures is often abstract and belongs to the realm of concepts and ideas. In
135
136 Julia Bamford
addition it can be used by lecturers to create rapport with the student listeners. They are able to do this partly because the vagueness of its referent enables a variety of meanings and associations to be attached to it. The fact that gestural reference is more common in lectures than in ordinary conversation, although only tentatively hypothesized in this paper, indicates that diŸerent genres use language in diŸerent ways and suggests that this could be a fruitful area for further research. Here is just one lexical item which is used diŸerently in diŸerent interactional circumstances. There are obviously many more including other deictics. The use of both small and larger corpora to make comparisons about the use of micro-features such as one particular spatial deictic also shows the need for further research which exploits the potential of comparing general corpora with a more specialized smaller corpus.
Notes 1. Recipient design, although conceptualised to explain features of spontaneous conversation, is particularly applicable to lectures where its qualities become accentuated. The organization and presentation of discourse is pre- prepared and includes features that are unlikely to be present in conversation, as, for example, when the lecturer announces his discursive intentions by giving an outline of what he is going to say. 2. This is linked to the discussion in Section 7 of symbolic uses of the deictic here which are used to convey emotion. 3. Fillmore’s original version of “Lectures on Deixis” was only available in mimeo form until the recent 1997 publication but precedes Levinson’s discussion. 4. Here occurs frequently in idiomatic expressions or pre-patterned language in conversation. In service encounters, for example, expressions such as ‘here you go’ often accompany handing over a cup of coŸee. 5. This obviously does not mean that listeners ªnd di¹culties is processing words or expressions with vague meanings. Channell’s experiments show in fact that listeners understand vague expressions such as ‘get me some oranges and things’ quite clearly as ‘get me 1 or 2 pounds of oranges and other prototypical fruits such as apples’.
References Bamford, Julia forthcoming “Interactivity in academic lectures: the role of question and answers”. In Dialogue within Discourse Communities , M. Bondi and J. Bamford (eds). Tübingen: Niemeyer.
Gestural and symbolic uses of the deictic here 137
2000
“Question and answer sequencing in academic lectures”. In Dialogue Analysis VII. Working with Dialogue, M. Coulthard, J. Cotterill and F. Rock (eds), 159–170 Tübingen: Niemeyer. Button, Graham and Lee, John R. E. 1987 “Transcript Symbols”. In Talk and Social Organization, G. Button and J. R. E. Lee (eds), 9–17. Clevedon: Multilingual Matters. Channell, Joanna 1994 Vague Language. Oxford: Oxford University Press. Fillmore, Charles J. 1997 Lectures on Deixis. Stanford: CSLI Publications. Glover, Kelly D. 2000 “Proximal and distal deixis in negotiation talk”. Journal of Pragmatics 32: 915–926. Goodwin, Charles 1981 Conversational Organization: Interaction between Speakers and Hearers. New York: Academic Press. 1996 “Transparent vision”. In Interaction and Grammar, E. Ochs, E. SchegloŸ and S. A. Thompson (eds), 370–404. Cambridge: Cambridge University Press. 2000 “Action and embodiment within situated human interaction” Journal of Pragmatics 32 :1489–1522. Halliday, M. A.K and Hasan, Ruqaiya 1976 Cohesion in English. London: Longman. Hanks, William F. 1990 Referential Practice: Language and Lived Space among the Maya. Chicago : Chicago University Press 1992 “The indexical ground of deictic reference”. In Rethinking Context: language as an interactive phenomenon, A. Duranti and C. Goodwin (eds), 43–76. Cambridge: Cambridge University Press. Hindmarsh, John and Heath, Christian. 2000 “Embodied reference: A study of deixis in workplace interaction”. Journal of Pragmatics 32: 1855–1878. Hyland, Ken 2000 Disciplinary Discourses. London: Longman. Levinson, Stephen 1983 Pragmatics. Cambridge: Cambridge University Press. Lyons, John 1977 Semantics. Volumes 1 and 2. Cambridge: Cambridge University Press. Matras, Yaron 1998 “Deixis and deictic oppositions in discourse: Evidence from Romani”. Journal of Pragmatics 29: 398–428. McCloskey, Donald 1985 The Rhetoric of Economics. Madison: University of Wisconsin Press.
138 Julia Bamford
Ochs, Elinor, Gonzales, Patrick and Jacoby, Sally 1996 ‘When I come down I’m in a domain state’: Grammar and graphic representation in the interpretive activity of physicists”. In Interaction and Grammar, E. Ochs, E. SchegloŸ and S. A.Thompson (eds), 328–369. Cambridge: Cambridge University Press. Quirk, Randolph, Greenbaum, Sidney, Leech, GeoŸrey and Svartvik Jan 1985 A Comprehensive Grammar of the English Language. London: Longman. Roth, WolŸ-Michael 2000 “From gesture to scientiªc language” Journal of Pragmatics 32: 1683–1714. SchiŸrin, Deborah 1990 “Between text and context: Deixis, anaphora, and the meaning of then” Text 10(3): 245–270. Simpson, Rita C., Briggs, Sarah L., Ovens, Janine and Swales, John M. 1999 The Michigan Corpus of Academic Spoken English. Ann Arbor, MI: The Regents of the University of Michigan.
The discourse function of contrastive connectors in academic abstracts Marina Bondi University of Modena and Reggio Emilia
1.
Introduction
The study of academic discourse has long proved to be a fruitful area of research for diŸerent perspectives on language, not only in the ªeld of EAP, but also in studies on writing, genre, text and discourse, to name but a few. The emphasis initially placed by genre studies on academic discourse as such has recently given way to an emerging interest in its cross-disciplinary variation. Hyland (2000), for example, emphasizes the need to consider the diŸerent sets of conventions and modes of inquiry that constitute each discipline and its “disciplinary culture”, with “a certain degree of interdisciplinary diversity and a degree of intradisciplinary homogeneity” (2000: 10). The study of abstracts may be particularly useful from this point of view, because of the central role played by abstract writing in expertto-expert communication. This paper presents some of the preliminary results of a comparative analysis of textual structures in abstract writing and focuses on the discourse functions of contrastive connectors as signals of the dialogic argumentative structures underlying abstracts in the human and social sciences. My emphasis is thus ªrst of all on areas of knowledge that diŸer from those of Hyland,1 but that roughly coincide with what he identiªes with the folk denomination of “soft” knowledge: economics, history and sociology. My analysis also starts from a slightly diŸerent perspective in terms of what I regard as the most important macro-functions of abstracts themselves. In line with Berkenkotter and Huckin’s view of abstracts as a promotional genre (1995: 34), Hyland emphasizes the need for abstract writers to typically situate themselves and their work in their disciplines, by displaying credibility and ‘membership’ (2000: 63). He studied the structure of abstracts using the standard IntroductionPurpose-Method-Results-Conclusion pattern2, though recognizing that the
140 Marina Bondi
schema of positivist enquiry hardly ªts humanities abstracts and is indeed rare in his corpus. His conclusions about the ‘soft’ disciplines highlight a double contrast: “while over 60 percent of abstracts in the soft disciplines contained introductions, this ªgure was only 30 per cent for the hard disciplines. The percentages were almost exactly reversed for methods” (2000: 70). He further notices that greater contextualization also re¶ects the diŸerent purposes of research: “writers in the softer ªelds frequently sought to discuss or deªne an issue, rather than establish empirical truths” (Hyland 2000: 72). My own emphasis is on the meta-discursive dimension of abstracts, the way they represent the typical discursive procedures of the discipline through the activity of the writer of the paper itself and of other members of the discourse community. This emphasis on the metadiscursive dimension of abstracts has already led me to associate the “representationality” of research article abstracts with the foregrounding and thematizing of main clauses referring to the development of discourse itself (Bondi 1997). A diachronic study of economics abstracts has allowed me to follow the historical development of content summaries into representations of the processes attributed to the author or the paper. The analysis of meta-argumentative expressions in particular (Bondi 1999) has highlighted how often scientiªc procedures are represented in terms of argument. The current paper explores the relationship between metadiscourse and speciªc disciplinary cultures in the use of connectors. Connectors and discourse markers in general,“ sequentially dependent elements which bracket units of talk” (SchiŸrin 1987: 31), have now long been the object of interest of linguists (SchiŸrin 1987, Blakemore 1987, Fraser 1990, 1996, 1998, Jucker and Ziv (eds) 1998)) in the analysis of both spoken and written discourse. Jucker and Ziv (1998: 1–2) emphasize the multiplicity of functions which discourse markers are said to fulªl and explicitly choose the convenient cover term of ‘discourse markers’ because of its fuzzy nature. Connectors are not simply seen in terms of the link they create between elements of propositional content, but also in terms of their interpersonal meaning, in line with approaches to the analysis of discourse that have constantly emphasized the interactive nature of text-structuring (Winter 1977, 1982; Hoey 1983, 1994, 2001). I will be looking at connectors, in the words of Sinclair (1993: 6), as “features which organize the sharing of meaning, as well as features which create the meaning”, as “part of the interactive apparatus of the language, progressively determining the status of a previous sentence in relation to the current one”, where coherence is realized by sequences of encapsulation and prospection. The very notion of ‘evaluative coherence’, the way in which “writers work to convey a consistent personal evaluation of the topic they are dealing with”, is used by Thompson and Zhou (2000: 123) to suggest that both attitudinal disjuncts and conjuncts may not be assignable to a single metafunction and that cohesion itself is an interpersonal as well as textual phenomenon.
The discourse function of contrastive connectors
Contrastive connectors in particular can be studied as signals of the dialogic argumentative structures underlying texts and foregrounded in abstracts. The view taken here is that they do not only enable monologic discourse to be interactive, but they also imply evaluation by assuming a common ground between reader and writer in terms of what is expected or unexpected at any given point in the discourse. Studies of contrastive connectors have often highlighted the problems of distinguishing semantic from pragmatic meaning in signals of contrast.3 SchiŸrin (1987), for example, studied the relationship signalled by but in terms of both ideational and interactional content.4 She concludes that the semantic meaning of but is contrastiveness and the pragmatic meaning is speaker-return, which has an expressive function of point-making. Her analysis of other contrastive connectors is limited to showing a dominant referential contrast in however and a dominant functional contrast in anyway. My intention is to study whether the notions of speaker-return and point-making could also be useful in an analysis of written texts, when these are studied from a dialogic perspective. This will be done through a case study of however in a corpus of abstracts, where anyway is totally absent and however seems to take over the conversational values of anyway. The complex classiªcations of contrastive markers in Fraser and MalamudMakowski (1996) and Fraser (1998) shift the interpretative emphasis from the speaker to the hearer. The hearer has to ªnd whether the contrast is with the direct or presupposed, entailed or implied message of the previous textual segment. Within this classiªcation, however is one of the most powerful contrastive connectors, representative of a major class of connectors that “signal that the speaker intends the explicit message conveyed by S2 to contrast with an explicit or indirect message conveyed by S1” (Fraser 1998: 306) as against those which signal correction (Fraser 1998: 307).5 My purpose, however, is not to study the diŸerences between however and other contrastive connectors, but rather to focus on a representative contrastive connector, in order to highlight its textual and discursive functions. Another important contribution to my analysis of contrastive connectors has been provided by argumentation studies, and in particular by the study of ‘l’argumentation dans la langue’.6 Some contrastive relations can best be explained with reference to the argumentative dimension of some common or shared knowledge called topos (Anscombre & Ducrot 1983). Topoi are part of our stereotyped encyclopaedic knowledge, which is shared by speakers of a language, as well as those sharing the same culture or way of thinking. Two tenets of this approach have been particularly useful: a.
the stereotypical functioning of meaning, i.e. the fact that language units are not explained in themselves but by the discursive chains they are associated with;
141
142 Marina Bondi
b. the dialogic dimension of meaning, i.e. the fact that any speaker presupposes a hearer with a similar or diŸerent point of view. Contrastive connectors have the special function to signal the need to disagree with a (hypo)thesis or to establish restrictions on it. Barton (1995) has already shown this from the point of view of an analysis of metadiscourse, where connective expressions are shown to function as interpersonal metadiscourse in written language as well as in spoken discourse, particularly in the language of academic argumentation. Barton analysed contrastive and non-contrastive connectors (those setting up an opposition and those explicitly denying opposition) and shows how they set up claims and counter-claims in a corpus of essays. She found that “contrastive connectives seem to be the preferred form of connective expression in claims” (Barton 1995: 230) and that the vast majority is realized in a two-part structure, with the presentation of a claim and an oppositional counter-claim. This has also been my conclusion in a previous study of dialogic patterns in academic textbooks (Bondi 1998). My purpose here is to explore how this analysis of connectors is re¶ected in the structure of journal article abstracts. The argumentative dimension of the genre lies not only in the anticipation of scientiªc claims to be established by the paper itself, but also in creating approval or interest in the reader as regards the purpose of the research/discussion. In order to elucidate the use of contrastive connectors in abstracts, I will start by exploring some frequency data illustrating the environments in which each connective occurs in each discipline, and I will then pay speciªc attention to the types of utterances/acts that directly precede or follow the connector, and to how they relate to the overall organization of the text in terms of moves (Sinclair and Coulthard 1975). In particular, I will investigate whether there is any disciplinary variation in the way contrast is used a.
to establish and emphasize the problematisation that characterizes introductions; b. to establish and emphasize a dialogic dimension in the results and conclusions presented.
2.
Materials and methods
The analysis is based on small corpora designed for the study of abstracts in diŸerent academic ªelds: economics, history and sociology. History was taken as the point of reference in the comparative analysis. The focus on history abstracts explains the smaller sizes of the other two corpora devised for the study. The
The discourse function of contrastive connectors 143
historical corpus contains abstracts for all journal articles published in 1994 in the abstracting service oŸered by Historical Abstracts. The comparative corpora include all the journal abstracts published in the ªrst issue of EconLit 1996 (1995 publications) and an equal number of journal abstracts selected from the Sociological Abstracts database (1995 publications). Table 1. The corpus of abstracts journal
year
number of abstracts
number of words
average number of words
Historical Abstracts EconLit Sociological Abstracts
1994
873
74,978
85.88
1996 1995
422 422
47,087 51,049
96.94 120.96
The analysis was carried out both from a quantitative and from a qualitative point of view. The ªrst step in the quantitative analysis was the setting up and exploration of frequency lists and key-words using Wordsmith Tools (Scott 1996). Wordsmith Tools identiªes key-words by comparing patterns of frequency. A word is identiªed as a positive or negative key-word if it occurs with unusually high or low frequency in the text as compared to its frequency in a larger reference corpus. Frequency wordlists and key-words — calculated using log likelihood — were explored in order to get a rough, preliminary picture of the respective role of contrastive connectors among connective items in general in the three disciplines under consideration. Use was also made of the frequency lists of the BNC Sampler, taken as roughly representative of a “general English” standard, and of the BNC Written corpus, for a “general written English” standard. The Sampler is a two percent sample of the full British National Corpus, selected from the full range of texts represented in the original corpus, about one million words spoken and one million words written. The Written BNC includes all the written texts in the corpus, irrespective of domain, time and medium; it thus includes both informative and imaginative texts: books, periodicals and miscellaneous. It is about ninety percent of the whole corpus, i.e. 90 million words. The second phase of the study was based on the analysis of the concordances of however in the three small corpora. The concordances were studied ªrst of all to isolate contrastive connectors (sentence adverbs, rather than subordinating conjunctions or intensiªers) and then to study basic patterns and meanings from a comparative point of view. The study of collocates and particularly of clusters in collocation proved particularly interesting with regard to highlighting cross-disciplinary variation.
144 Marina Bondi
In the ªnal stage, the historical abstracts corpus was subjected to a more indepth textual analysis of the use of however in order to identify the core-meanings of the connector and the textual patterns in which these meanings were found.
3.
Results and discussion
3.1 Key-words and frequency lists A very rough and very preliminary overview of the role of contrastive connectors among connectors in general can be obtained by scanning the wordlists of my corpora and comparing them to some kind of non-homogeneous corpus. For this preliminary overview, I have chosen a few easily identiªable one-word connectors: although, but, however, if, or, so, then, when, while. Quantitative data, of course, do not distinguish the speciªc meanings of the forms under consideration, and the need to integrate a consideration of frequency data (word-lists and key-words) with a closer study of contextual data by means of concordances could hardly be overestimated.7 Table 2 shows selected data from my three corpora and from the BNC, for comparison. The table includes the number of all occurrences of each connector in my small corpora, in the BNC Sampler and in the Written BNC. For comparative purposes the number of occurrences per ten thousand words is also included in brackets. Table 2. Frequency counts of selected connective items (Numbers in brackets represent frequency per 10,000 words) connector
historical abstracts
econlit
sociological abstracts
BNC sampler
written BNC
although but however if or so then when while
55 182 89 18 120 36 21 54 82
16 59 42 77 93 9 39 67 32
48 88 30 12 102 12 14 34 60
584 10,626 700 6,742 6,968 7,782 5,007 4,708 836
4,2032 395,383 59,648 213,423 341,303 189,710 124,729 187.831 54,195
(7) (24) (12) (2) (16) (5) (3) (7) (11)
(4) (14) (10) (19) (22) (2) (9) (16) (8)
(10) (18) (6) (2) (21) (2) (7) (7) (12)
(3) (50) (3) (32) (33) (37) (24) (22) (4)
(5) (44) (7) (24) (38) (21) (14) (21) (6)
Table 3 reports the key-ness indexes provided by Wordsmith Tools when comparing my small corpora to the BNC Sampler. Shaded areas show positive key-words, i.e. words occurring with unusually high frequency. All other indexes reported refer to negative key-words, i.e. words occurring with unusually low frequency.
The discourse function of contrastive connectors 145
Table 3. Key-ness index of selected connective items in the three small corpora (Reference Corpus: BNC Sampler) connector
econlit
sociological abstracts
although but however if or so then when while
---−187.0 +29.8 −40.6 −26.0 −264.8 −61.3 -------
+50.5 −134.5 ---−225.1 −25.2 −269.6 −143.6 −70.3 +52.3
historical abstracts +36.6 −123.7 +92.8 −344.8 −80.1 −329.2 −220.5 −102.6 +58.6
The data seems to conªrm that abstracts tend to use a limited number of connectors in general. Most items, in fact, have lower frequencies in my corpora than in the BNC Sampler corpus and often turn out to be negative key-words. If it is true that the abstracts of my specialized small corpora, on the whole, tend to use fewer connective elements than the general English texts included in the BNC Sampler, it cannot be denied that some connectors show much higher frequencies. Two of these are typically contrastive connectors: although and however. While is also basically contrastive, when concordances are looked into. Cross-disciplinary variation shows subtler patterns and highlights interesting data that seem to ªnd an explanation in the epistemology of the disciplines. Economic abstracts, for example, show a noticeably higher frequency of if and when. If is also highlighted as a key-word of the economics corpus when studied with reference to the other two corpora (key-ness index 94.0). This may be interpreted as re¶ecting the typical forms of reasoning of the discipline, with its characteristic forms of hypothetical reasoning based on mathematical models. A closer look at the concordances of when would also conªrm that its predominant use is not temporal, but rather conditional. As far as history is concerned, comparison with general English highlights a much higher frequency of however (key-ness index 92.8) and while (key-ness index 58.6), with while predominant in its contrastive, rather than temporal, meaning. Cross-disciplinary comparison also highlights another element that may be very important for the deªnition of the generic structure of these abstracts: one of the most signiªcant key-words is reviews with a percentage frequency of 0.11% as against 0.01% in the economic and sociological abstracts together (key-ness index 79.3). This diŸerence may be interpreted as suggesting that the frequency of review articles themselves is generally much higher in historical journals and above all in historical abstracting services, when compared to socio-economic disciplines. This
146 Marina Bondi
does not just remind us of the importance of historiography within historical studies, but also has important implications for the deªnition of a generic structure of articles and of the abstracts themselves. One of the guiding principles for the analysis has thus been the idea that contrast plays a major role in the use of sentence connectors made by journal article abstract writers. However in particular plays a key role in historical abstracts. This prominence of however in historical journal article abstracts is also shown by comparing its frequency in my corpus of journal article abstracts — 12 per 10,000 words — with its frequency in the whole database of 1994 Historical Abstracts, including abstracts of reports and books: 7 per 10,000 words. The role played by however in signalling patterns of macro-connectivity is shown by concordances to be dominant in comparison with other potential contrastive connectors: actually (even)/(al)though, but, conversely, despite, even so, in comparison, in contrast /by contrast, in fact, in reality, in spite of, instead, nevertheless, nonetheless, on the contrary, on the other hand, rather, still, whereas, while, yet. Even if however is not the most frequent connective in absolute terms, it is dominant for sentence connection (connection above the clause-complex), whereas but and while dominate in intra-sentential connection (within the clause-complex). Considering historical abstracts alone, for example, a mere two of the 89 occurrences of however are subordinating conjunctions whereas 87 are conjunct adverbs. But and while, of course, show reverse patterns; while is exclusively used as intrasentential conjunction, whereas but can in fact be used in both contexts: only 13 of its 182 occurrences are sentence connectors, whereas 167 are conjunctions (at various levels) within the clause-complex. The other two occurrences are examples of a preposition (meaning ‘except for’) and an adverb meaning ‘only’.
3.2 Meaning, scope and position of however The contextual meaning of the various occurrences of however in my corpora seems to conform to SchiŸrin’s analysis of but: not only is the semantic meaning ‘contrastiveness’, but also the pragmatic meaning seems to be some kind of ‘speakerreturn’, with an expressive function of ‘point-making’. Given the basically argumentative nature of academic communication, it seems justiªed to resort, for a more speciªc deªnition of the meaning of however, to the notion of argumentative role as “la fonction, oŸensive ou défensive, que la phrase est capable d’exercer dans le mécanisme de la persuasion: preuve, rectiªcation, conclusion, etc.” (Stati 1990: 16).8 From the point of view of an analysis of argumentative roles, however signals the presentation of a claim as Counter-claim: the connector highlights the Writer’s stance (Discourse) against the background of its Counter-discourse. The contrast
The discourse function of contrastive connectors 147
can be both with the explicit message of what precedes and with any of the indirect messages that are associated with the textual section encapsulated by however. The pragmatic meaning of ‘speaker-return’ accompanies what is argumentatively a shift from a Concession to a (Counter)-Claim, Concession^(Counter)-Claim (1) Given the changing nature of historical materials and of the new questions each generation asks, history is inevitably unªnished. However even as victims of amnesia would not be expected to handle their own aŸairs, people ignorant of their historical past cannot understand their own present. The status of the Claim within the global textual structure may diŸer widely in the various occurrences of the connector. The signal itself, however, is still eminently dialogic: it encapsulates one kind of discourse and contrasts it with another, thus signalling Writer’s stance. When analysing a genre like the abstract, so explicitly dominated by the rules of economy and by the need to be brief and concise, deciding on the scope of a connector can easily lead the analyst from issues of micro-connectivity to issues of macro-connectivity. The scope of however in the specialised history corpus easily encompasses large sections of an abstract, occasionally constituting the basic structure of the abstract itself, as in (1) above or in (2) below. (2) The Phelps-Stokes educational proposals for South Africa, which stressed vocational training, concentration on rural schools and self-help, were exactly what the African population needed early in the 20th century. These practical proposals were, however, defeated by local white interests, who wanted an unskilled labor force, and by African intellectuals, who wanted to expand elite opportunities. Position of the sentence adverb may also turn out to be an interesting object of study. In my corpus of historical abstracts it is mostly in initial position (38/87) or mid-position (47/87). Examples of ªnal position are very limited (2/87), but interestingly linked to patterns of anticipation: the clause complex concluded by however actually contains other elements anticipating that further speciªcation of informationally new elements is needed, as in example (3). (3) In newly reunited Germany in the early 1990’s, right-wing extremism, the electoral success of the new right-wing parties, and the existence of a “party of non-voters” disillusioned with established politics oŸered disquieting parallels with the early 1930’s. The economic and political threats to Weimar democracy were largely missing sixty years later, however. German national and regional elites in the early 1990’s were committed to democ-
148 Marina Bondi
racy, although reluctant to deal energetically with social problems, such as assimilation of immigrants and refugees, that aggravated extremism. Mid-position seems to divide the clause into two parts. If the second is more clearly identiªed as the informative focus of the message (the ‘new’ element), the ªrst is not necessarily explicitly ‘given’ as in (2) above. Extract (4) shows, for example, a case where the ªrst element of the clause is another signal of textual pattern, showing a movement from speciªc to general consideration. In Hallidayan terms the two elements would be multiple themes, i.e. the textual (overall) and the interpersonal theme (however) (Halliday 1985: 53–54): (4) Examines the gross domestic product (GDP) of the Soviet Union and the East European countries in 1937–38 in order to estimate how far they lagged behind the Western industrial nations. Analysis based on purchasing power parity, exchange rates, physical indicators, and multiple regression methodology shows a wide range of estimates and reveals a number of variables. Overall, however, the ªndings support the observation of Joseph Stalin (1879–1953) that in the 1930’s the Soviet Union lagged behind the industrialized nations by ªfty to a hundred years. It is also clear that Czechoslovakia and Hungary were the most developed of the East European nations, with Bulgaria, Poland, Romania, and Yugoslavia at similar low levels of development. On the whole, the elements that precede however seem to play some kind of textstructuring function: they may just consist of (or include) an encapsulator (as shown in examples 2 and 5), but they may also contain a signal of a shift in theme, whether ideational (examples 7 and 8) or textual (example 4). The latter is perhaps the most interesting case from the point of view of the macro-structure of abstracts. It highlights the importance of the elements that accompany however, which can also be identiªed by studying collocates and clusters of words in the immediate context of the connector. The small corpus of historical abstracts does not show signiªcant features of lexical repetition. It is only when we study wider sets of data that patterns emerge. Concordances of however in the complete historical abstracts data-base of journal article abstracts for the 1982–1996 period (442,086 words) yield 10,949 examples of however. An analysis of clusters in these concordances shows a decided preference for time adverbials (In the 20th century, at the end of the 19th century, between June and September, at the same time, from 1939 to 1945, by the 20th century, in the early/ late forties/eighteenth century…). The quantitative predominance of these clusters shows that however mostly accompanies signals of a narrative structure. Other signals, like a result of, in the case of, a close examination of, it is clear that, it is possible
The discourse function of contrastive connectors 149
to, there is no evidence…, show a closer link with causal patterns and the argumentative structure of abstracts. Cross-disciplinary study of collocates and clusters involving however oŸers further useful information. In the small corpus of economics abstracts, on the one hand, the connector is mostly associated with the meta-argumentative verb ªnd (we also ªnd that…). The small corpus of sociological abstracts, on the other hand, shows closely associated clusters like it is argued, it is suggested, it is concluded. In abstracts from the ªelds of economics and sociology, therefore, the connector shows a clearer association with other signals of point-making that represent the experimental process of inquiry.
3.3 The discourse function of contrastive connectors in historical abstracts The most important meta-pragmatic function of contrastive connectors is thus to act as signals of the dialogic argumentative structures underlying abstracts. This function plays a major role both at the level of the macro-functions of the genre, macro-functions like Hyland’s (2000) ‘claiming signiªcance’ and ‘claiming credibility’, and at the level of the generic structure of the connectors. Using the most obvious of the subdivisions within the genre, and one that is signalled as a particularly important distinction by the unexpectedly high frequency of the verb reviews (see discussion of key-words above), one should probably start by clearly distinguishing the abstracts of review articles form those of research articles. In abstracts of review articles, such as example (5), however highlights the dialectic nature of reviewing; it marks the shift from positive to negative features, in terms of argumentative roles: from Agreement/Praise to Disagreement/Criticism: (5) Reviews nine books which reveal a liberal consensus that the spread of democracy, free trade, and the growth of interdependence is producing an unprecedented era of peace and cooperation in inter-American relations. This perspective, however, does not provide clear answers to problems growing out of US power or the costs connected with neoliberal models. In abstracts of research articles, the discourse function varies according to the epistemology highlighted by the speciªc abstract. As I have discussed elsewhere (Bondi in press), this time the diŸerence is not in typological terms (either/or), but rather ‘topological’, a matter of showing fewer or more characteristics of two not necessarily exclusive ‘prototypes’: abstracts that tend to emphasize the notion of “History as story-telling, or narrative” and those that tend to emphasize the notion of ‘history as argument’. The variation can be indicated on a cline as shown in Figure 1:
150 Marina Bondi
__________ ___ ___ __ __ _ _ _ _ __ __ __ ___ ___ ___________ History as narrative History as argument
Figure 1. Topological variation in historical abstracts
In abstracts where emphasis is placed on the narrative element of history, the text tends to coincide with sequences of sentences reporting causally and temporally linked processes, and however contributes to claiming signiªcance and credibility by: a.
highlighting the “problematicity” of an initial situation to be explained → Problematizing (Example 6) b. showing the unexpectedness of an ending/explanation (“resolution”) → Claiming signiªcance (Example 7) The distinction is not always clear-cut, especially when articles represent events synthetically. The distinction, however, can be exempliªed in (6) and (7) below: (6) Between 1750 and 1850 Indian armies such as those of Mysore, the Marathas, and the Sikhs made signiªcant strides in military organization. However, they were defeated by the forces of the East India Company because the British, aided by a vastly superior organizational and governmental infrastructure and a strong geographic position in India, were able to adapt more rapidly to the changing military situation in India. By anticipating and preempting the advances made by their Indian opponents, the British thwarted any decisive actions that could be made against them. (7) Examines the evolution of teacher training in Papua New Guinea, 1873– 1985. Much of its history is tied to the history of the Christian missions. The period 1873–1946 was one of classical colonialism, but a second era commenced after World War II characterized by the philosophy that social change in developing countries should be brought about slowly and that a democratic society should be based on universal primary schooling. In 1958–71 an expansion of the primary school system took place, and after independence in 1975 there was a major national policy shift toward decentralization; teacher training, however, remained under national control.
In abstracts that more explicitly refer to ‘history as argument’, whether following empirical models (such as the Introduction-(methods)-results-discussion pattern) or dialectical models of arguing by balancing diŸerent opinions, however can be found to contribute to claiming signiªcance and credibility by
The discourse function of contrastive connectors
a.
showing the novelty of an issue → Problematizing (Claiming novelty) (Example 8) b. relating the claim to debate within the discourse community → Claiming signiªcance (Example 9) c. highlighting “incoherence” in evaluation of results/ data/ conclusions → Signalling stance (Example 10) (8) Talk of crisis and transition is extremely common in discussions of the industrial restructuring processes that occurred in the Western world in the 1980’s and early 1990’s. Not enough attention has been paid, however, to the rhetoric of crisis, or its ability to invoke and not simply describe a rite of passage. Crises do not occur naturally; they are culturally constructed through narrative. The article looks, therefore, at the politics of crisis-talk, especially at the way this talk rewrites the story of the time leading to the newly recognized crisis. While crisis-talk naturalizes certain political trajectories by aligning them with the future and therefore maturity and advancement, it delegitimates other political projects by aligning them with the past and therefore with immature and underdeveloped conditions. Moreover, it does this in a liminal atmosphere of political panic. Newcastle, Australia, was a muse through which Australians understood what was happening to national industry and the nation more generally in the 1980’s. (9) Historical comparisons of the Soviet and US economies most often focus on the stark contrasts between market and centrally controlled systems. However, expediency in response to dynamic pressures caused the adoption of central planning in both countries. After summarizing Soviet central planning experiences before 1965, similarities to US wartime planning are discussed. These comparisons provide insights into the problems of moving between market and planned economies. (10) Describes the author’s experiences in transcribing and checking the oral history interviews he conducted with Sir Laurence Hartnett, chief executive of General Motors-Holden, 1934–47. Hartnett’s testimony was found to contain substantial inaccuracies, of varying degrees of historical importance. However, this should not be assumed to lessen the value of oral history. Examination of the reasons underlying the inaccuracies leads to a more complete picture. If it is sometimes di¹cult to classify moves precisely as one or the other, especially in shorter abstracts, the function of however remains clear. Whatever the sub-genre, the macro- and the micro-function, however helps to establish and emphasize a
151
152 Marina Bondi
dialogic dimension in the problems, results and conclusions presented. And it greatly contributes to the constitutive “representationality” of abstracts, i.e. their meta-discursive references to the discursive procedures of the original article as objects of abstracting discourse, thus oŸering a discipline-speciªc representation of scientiªc procedures. A parallel exploratory study of however in the corpora used for comparison highlights for example a much more clear-cut distinction in the economics corpus between two basic types of abstracts: those following the pattern of model-testing (thus presenting claims as logically derived from correctly applied procedures) and those highlighting the role played by dialectic reasoning in presenting an argument before the discourse community. However mostly signals unexpected results or the possibility of diŸerent interpretations of the results as well as cautious conclusions. Economists seem to prefer to establish argumentative dialogue on the basis of the interpretation of their results rather than on the novelty of the issue dealt with. Similarly, sociological abstracts seem largely based on the process of inquiry, and however highlights the novelty and credibility of the results mainly by pointing out their signiªcance (70%) and only more rarely by constructing dialogue within the discourse community (20%).
4.
Conclusions
This paper has provided a brief analysis of the ways in which contrastive markers signal the argumentative roles played by textual units in the structure of abstracts. Much more could be said about the details and complexities involved in the use of contrastive markers in academic discourse in general. The study has focused on a case study of however as a contrastive connector in historical abstracts. This admittedly limited point of view has helped in the analysis of the core meaning of contrastive expressions and their role in speciªc generic structures and disciplinespeciªc argumentative procedures. Perhaps most importantly, the ªndings presented here show the need to consider multiple dimensions of language variation in the analysis of discourse patterns and their markers. The interplay between text and context can be analysed with diŸerent degrees of delicacy, with reference to both discourse, deªned as the general ªeld of social activity in which the speech event takes place, and to genre, deªned as the class of communicative events to which the speciªc set of communicative events being examined belongs. The meta-argumentative dimension of contrastive connectors can be seen as constitutive in the deªnition of both academic discourse in general and academic abstracts in particular.
The discourse function of contrastive connectors
In addition to highlighting the importance of re¶exivity and argumentation in academic discourse, the study has shown the close link between language choice and epistemology in academic discourse. When considered from the point of view of the wider study of metadiscourse across disciplines, the case study of however conªrms the idea that metadiscursive expressions highlight issues that re¶ect the epistemological ethos of the disciplines, i.e. the way they represent their own activities and procedures as against major methodological debates. Hopefully, the study has also contributed to showing the need for corpusbased analyses to complement more intensive studies of particular texts, and vice versa. Text and discourse studies can only be fully developed when closer analysis of particular instances of communicative events is integrated with quantitative data from wider textual bases. These provide the necessary background and support to textual interpretation.
Notes 1. Hyland (2000) is based on data from the following areas: Philosophy, Sociology, Applied Linguistics, Marketing, Biology, Electronic Engineering, Mechanical Engineering and Physics. 2. Dos Santos (1996) starts from Swales’ (1990) description of the generic structure of abstracts, based on a sequence of Introduction, Methods, Results, Discussion, with the Introduction establishing the ªeld, summarizing previous research, showing a gap, introducing current research, and proposes the following structure: 1. Situating the research (Stating current knowledge; Citing previous research; Extending previous research; Stating a problem). 2. Presenting the research (Indicating main features; Indicating main purpose; Hypothesis raising). 3. Describing the methodology. 4. Summarizing the results. 5. Discussing the research (Drawing conclusions; Giving recommendations). 3. See for example LakoŸ (1971), Halliday and Hasan (1976), Dascal and Katriel (1977), SchiŸrin (1987), Fraser and Malamud-Makowski (1996), Park (1998), Fraser (1998) and Takahara (1998). 4. SchiŸrin’s study of but distinguishes its use for referential contrast, functional contrast and contrastive actions, i.e. “speakers’ eŸorts to make a point in reaction to interruptions, distractions, challenges and disagreements” (SchiŸrin: 1987: 177), but she also admits that the connector can mark all of three types of contrast simultaneously. 5. Fraser and Malamud-Makowski (1996: 880) suggest that both but and however signal denial of a proposition arising from the previous discourse, but they also admit that however, when compared to but, brings a sense of concession (Fraser and MalamudMakowski 1996: 868). 6. See the special issue of Journal of Pragmatics 1995 (24, 1–2), dedicated to “Argumentation within the language theory”.
153
154 Marina Bondi
7. The relatively low frequency of but, for example, can only be explained with reference to a closer study of its variation. Even if a signiªcant number of instances in the Sampler corpus could be examples of conversational use of ªnal but (eliciting Hearer’s inference), this feature alone does not justify the diŸerence, which is also shown between the BNC written corpus and my specialized small corpora (Table 2). 8. This may be distinguished from the illocutionary force of a speech act, although both functions pertain to the area of pragmalinguistics, and, in their re¶exive dimension — to metapragmatics (see Verschueren 1995; Lucy, 1993: 11–21; Silverstein 1993: 33–35). In classifying argumentative roles, Stati (1990: 65–85) considers the following categories: (a) ‘assentiment, conªrmation, adhésion’; (b) ‘justiªcation, preuve’; (c) ‘concession’; (d) ‘rectiªcation’; (e) ‘objection’; (f) ‘contestation, désaccord, dissentiment’; (g) ‘critique, accusation, reproche’.
References Anscombre, Jean-Claude and Ducrot, Oswald 1983 L’argumentation dans la langue. Bruxelles: Mardaga. Barton, Ellen 1995 “Contrastive and non-contrastive connectives: Metadiscourse functions in argumentation”. Written Communication April: 219–239. Berkenkotter, Carol and Huckin, Thomas 1995 Genre Knowledge in Disciplinary Communication: Cognition / culture / power. Hillsdale, N. J.: L. Erlbaum. Blakemore, Diane 1987 Semantic Constraints on Relevance. Oxford: Blackwell. Bondi, Marina 1997 “The rise of abstracts: Development of the genre in the discourse of economics”. Textus, X, E. Barisone and G. Hughes (eds): 395–418. 1998 “Dialogues within discourse communities in economics textbooks”. In: Dialoganalyse VI, S. Cmejrková, J. HoŸmannová, O. Müllerová and J. Svetlá (eds), 229–238. Tübingen: Niemeyer. 1999 English Across Genres: Language variation in the discourse of economics. Modena: Il Fiorino. In press “Metadiscursive practices in academic discourse: Variation across genres and disciplines”. In: Dialogue within Discourse Communities: Metadiscursive perspectives on academic genres, M.Bondi and J.Bamford (eds), 3–28. Tübingen: Niemeyer. Cmejrková, Svetla, HoŸmannová, Jana, Müllerová, Olga and Svetlá, Jindra (eds) 1998 Dialoganalyse VI. Tübingen: Niemeyer. Dascal, Marcelo and Katriel,Tamar 1977 “Between semantics and pragmatics: The two types of ‘but’ — Hebrew ‘aval’ and ‘ela’”. Theoretical Linguistics 4: 143–172.
The discourse function of contrastive connectors
Dos Santos, Mauro Bittencourt 1996 “The textual organization of research paper abstracts in applied linguistics”. Text 16(4): 481–499. Fraser, Bruce 1990 “An approach to discourse markers”. Journal of Pragmatics 3: 383–395. 1996 “Pragmatic markers”. Pragmatics 6(2): 167–190. 1998 “Contrastive discourse markers in English”. In Discourse Markers, A.Jucker and Y.Ziv (eds), 301–326. Amsterdam/Philadelpha: John Benjamins. 1998 “What are discourse markers”. Journal of Pragmatics 31: 931–952. Fraser, Bruce and Malamud-Makowski, Monica 1996 “English and Spanish contrastive discourse markers”. Language Sciences 18(3–4): 863–881. Halliday, M. A. K. and Hasan, Ruqaiya Cohesion in English. London: Longman. Halliday, M. A. K. 1985 An Introduction to Functional Grammar. London: Arnold. Hoey, Michael 1983 On the Surface of Discourse. London: Allen and Unwin. 1994 “Signalling in discourse: a functional analysis of a common discourse pattern in written and spoken English”. In Advances in written text analysis, M. Coulthard (ed.), 26–45. London: Routledge. 2001 Textual Interaction: An introduction to written discourse analysis. London: Routledge. Hyland, Ken 2000 Disciplinary Discourses: Social interaction in academic writing. London: Longman. Jucker, Andreas and Ziv,Yael (eds) 1998 “Discourse markers: Introduction”. In Discourse Markers. Description and Theory, A. Jucker and Y. Ziv (eds), 1–12. Amsterdam/Philadelphia: John Benjamins. LakoŸ, Robin 1971 “If’’ s, and’s and but’s about conjunction”. In: Studies in Linguistic Semantics, C. J. Fillmore and D. T. Langendoen (eds), 115–150. New York: Holt, Reinhart, and Winston. Lucy, John 1993 “Re¶exive language and the human disciplines”. In Re¶exive Language. Reported Speech and Metapragmatics, J. Lucy (ed.), 9–32. Cambridge: Cambridge University Press. Park, Yong-Yae 1998 “A discourse analysis of contrastive connectives in English, Korean, and Japanese Conversation: With special reference to the context of dispreferred responses”. In Discourse Markers. Description and Theory, A.Jucker and Y. Ziv (eds), 277–300. Amsterdam/Philadelpha: John Benjamins.
155
156 Marina Bondi
Redeker, Gisela 1991 “Review article: Linguistic markers of discourse structure”. Linguistics, 29 (6): 1139–1172. SchiŸrin, Deborah 1987 Discourse Markers. Cambridge: Cambridge University Press. Scott, Michael 1996 Wordsmith Tools. Oxford: Oxford University Press. Silverstein, Michael 1993 “Metapragmatic discourse and metapragmatic function”. In Re¶exive Language. Reported Speech and Metapragmatics, J. Lucy (ed.), 33–58. Cambridge: Cambridge University Press. Sinclair, John 1993 “Written discourse structure”. In Techniques of Description, J. Sinclair, M. Hoey and G. Fox (eds), 6–31. London: Routledge. Sinclair, John and Coulthard, Malcolm 1975 Towards an Analysis of Discourse. Oxford: Oxford University Press. Stati, Sorin 1990 Le Transphrastique. Paris: PUF. Swales, John 1990 Genre Analysis: English in academic and research settings. Cambridge: Cambridge University Press. Takahara, Paul Osamu 1998 “Pragmatic functions of the English discourse marker anyway and its corresponding contrastive Japanese discourse markers”. In Discourse Markers. Description and Theory, A. Jucker and Y. Ziv (eds), 327–351. Amsterdam/ Philadelpha: John Benjamins. Thompson, GeoŸ and Zhou, Jianglin 2000 “Evaluation and organization in text: The structuring role of evaluative disjuncts”. In Evaluation in Text: authorial stance and the construction of discourse, S. Hunston and G. Thompson (eds), 121–141. Oxford: Oxford University Press. Verschueren, Jef 1995 “Metapragmatics”. In IPrA Handbook of Pragmatics, J. Verschueren, J.-O. Östman, J. Blommaert and C. Bulcaen (eds), 367–371. Amsterdam/Philadelphia: John Benjamins. Winter, Eugene 1977 “A clause-relational approach to English texts: A study of some predictive lexical items in written discourse”. Instructional Science 6: 1–92. 1982 Towards a Contextual Grammar of English. London: Allen and Unwin.
The discourse functions of I don’t know in English conversation* Giuliana Diani University of Modena and Reggio Emilia
1.
Introduction
It is widely accepted that conversation involves us in doing things, in getting others to do things, in eliciting information, in refusing, apologising, promising and threatening, and in a whole range of other activities. My concern is with investigating why people respond in the way they do and in particular how they use I don’t know in conversational exchanges. There is a general expectation that speakers cooperate and use language to facilitate the conveyance of information. This is not the case for I don’t know, which does not seem to be used with its literal meaning. What I want to do in this paper is analyse the meanings of I don’t know in face-toface conversations between native speakers of English and investigate the motivation behind its production and the pragmatic functions that it performs. I don’t know is generally taken to be a reply to an information question when the speaker is unable to supply the requested information. However, it is a common experience that I don’t know is produced even when the speaker is able to supply the information. Her/his declaration of insu¹cient knowledge can be explained as the result of attending to participants’ face (GoŸman 1967). Politeness can be seen as maintaining participants’ willingness to communicate, thus allowing information transfer to take place. My study suggests that the pragmatic motivation behind the production of I don’t know is often a concern to respect principles of politeness.1 In common parlance we talk of acting politely, where politeness is a means to maintain rapport while getting on with the business at hand. It makes communication more smooth because if the hearer is oŸended s/he may walk out or never consent to speak to her/his interlocutor again. As Widdowson (1983: 48) puts it: It is all very well to believe in being blunt and plain-spoken, but to be so is to rely on a tolerance that many people might be reluctant to extend, and they are likely, in this case, to disengage from the interaction.
158 Giuliana Diani
Very little work has been done on the topic which is the focus of this paper, the discourse functions of I don’t know. As far as I know, only Tsui (1991) has shed some light on this subject. In her study, she oŸers some useful insights into the pragmatic functions of I don’t know by arguing, for example, that I don’t know occurs in conversational environments other than in reply to information questions and that it has a variety of functions in addition to a declaration of inability to supply information. Tsui’s description has provided a useful starting-point for analysing the diŸerent functions of I don’t know. In my paper I have gone one step further and investigated how certain ‘discourse markers’ (SchiŸrin 1987) occurring in conjunction with I don’t know aŸect its function. The present paper is divided into three sections. In Section 1, I discuss GoŸman’s (1967) theory of ‘face’ and Brown and Levinson’s (1987) notion of politeness which, I argue, provide a conceptual framework which is appropriate for analysing the functions of I don’t know. In Section 2, I begin by presenting the occurrences of I don’t know which are normally regarded as prototypical, such as the reply to information questions; I then go on to examine how I don’t know realizes diŸerent pragmatic functions. In the third section, I consider the role of certain discourse markers occurring in conjunction with I don’t know, examining how they in¶uence its function. The data used are taken from the COBUILD/Birmingham Spoken Corpus, which consists of about two million words and is composed of 133 texts representing recordings of everyday casual conversation, meetings, interviews and discussions. In the corpus there were 1,114 occurrences of I don’t know.2 The occurrences were studied in order to: – – –
2.
explore the potential value of the notions of ‘face’ and politeness in the interpretation of I don’t know; identify the meaning and functions of I don’t know; study the collocations of I don’t know with certain discourse markers on the basis of their pragmatic meanings.
The notions of ‘face’ and politeness
When they engage in talk exchange people tend to show deference, ‘respect’ for each other, a behaviour compatible with ‘face’ (GoŸman 1967), i.e. the public selfimage people want to claim for themselves. GoŸman sees face as a negotiated outcome of talk, as something which can be strategically obtained. He deªnes face as “the positive social value a person eŸectively claims for himself by the line others assume he has taken during a particular contact” (1967: 5). Thus, face is something
The discourse functions of I don’t know 159
that can be lost, maintained, or enhanced, and must be constantly attended to in interaction. In general, people cooperate (and assume mutual cooperation) in maintaining face in interaction. Respect for the other implies that expressions of disagreement and of diŸerences of opinions are controlled or ‘softened’ so as to minimize the face-threatening eŸect of a negative assessment. Therefore, instead of providing a dispreferred response to a preceding assessment, that is disagreement, the speaker may employ protective manoeuvres so that the other’s face is preserved. This use of I don’t know as a face-saving mechanism is illustrated in example (1): (1) <M05>
yeah she put put them back in their place and and and made them more reasonable. <M01> Well I don’t know. Er I think that people like erm Scargill needed to be chopped down and she chopped him down. (Cobuild: ukspok/04. Text: S000000017)3
Instead of producing an assessment immediately, which might con¶ict with M05’s statement, M01 responds by ªrst saying I don’t know with the result that the subsequent assessment is pushed forward into the turn. The utterance I don’t know, then, signals a con¶ict between the two turn components. On the one hand, M01 appears to claim insu¹cient knowledge of the topic, but on the other hand, by proŸering a second assessment immediately afterwards, he is in fact claiming knowledge of the topic. M01’s initial declaration of insu¹cient knowledge seems to be a matter of attending to M05’s face. Examples of this kind are frequent in the corpus. What people really want in interaction is to establish ‘a state where everyone temporarily accepts everyone’s line’ (GoŸman 1967: 11). Sometimes, to avoid contradiction, the speaker leaves unstated facts that might implicitly or explicitly contradict and embarrass the positive claims made by others. The speaker may proŸer her/his replies with careful ambiguity so that the other’s face is preserved. This is illustrated in example (2): (2) […] that can do a really good job like Maggie Thatcher. <M01> Now there seems to be I mean I don’t know this is not scientiªc at all but a bit of a sexual divide here. It seems the women like […] (Cobuild: ukspok/04. Text: S9000000810)4 Here I don’t know can be explained according to GoŸman’s (1967: 15) ‘avoidance process’. Avoiding expressing one’s own opinion is the surest way for a person to prevent threats to her/his own face. I don’t know is not motivated by the speaker’s inability to provide the information required, but by his desire to reduce his commitment towards what he is saying.
160 Giuliana Diani
Like GoŸman, Brown and Levinson (1987) are concerned with the concept of face. Proposing a modiªed version of GoŸman’s deªnition of face, as “the public self-image that every member wants to claim for himself” (1987: 62), they note that some speech acts are intrinsically face-threatening acts (FTAs). In consequence, unless S’s want to do an FTA with maximum e¹ciency is greater than S’s want to preserve H’s (or S’s) face to any degree, then S will want to minimize the face threat of the FTA. (Brown and Levinson 1987: 60)
Such attention to face is clearly an interactional concern — it relates to the maintenance of rapport. Brown and Levinson see the e¹ciency of information transfer as reduced by the need to simultaneously achieve the interactional goal of attending to face. Attending to participants’ face is a matter of acting politely, by counteracting “the potential face damage of the FTA by doing it in such a way, or with such modiªcations or additions, that indicate clearly that no such face threat is intended or desired” (Brown and Levinson 1987: 69–70). This data shows that a high degree of politeness and co-occurrence with a large number of mitigation strategies characterize many instances of I don’t know. In the vast majority of cases, the speaker uses I don’t know to reduce her/his commitment to the truth of the proposition expressed, hence mitigating the face-threatening eŸect of a negative assessment, as examples (3) and (4) illustrate: (3)
[…] if somebody had done it in car. It doesn’t seem fair. Whether it’s right or wrong I don’t know. <M01> Do you think it’s harsher or not as harsh?
(4) […] that point of view and I was not very impressed with it. Er [laughs] you may feel the same I don’t know but that’s that’s part of the point of of investigating these positions […] (Cobuild: ukspok/04. Text: S0000000427) By saying I don’t know, the speakers seem to open up the possibility that their statements may be incorrect. The speakers seem to advise their co-participants that they are just expressing a personal opinion. The point I want to make is that I don’t know suggests either that the speaker is not really sure of what s/he is saying, or s/he does not want to take full responsibility for the truth of her/his utterance, so as to neutralize a potentially oŸensive act (cf. R. LakoŸ 1973a). We see that there may be diŸerent motivations for the choice of a linguistic expression such as I don’t know. In the remainder of the paper, I will discuss the major functions that I don’t know performs in discourse, adopting, as a framework, the analysis by Tsui (1991).
The discourse functions of I don’t know
3.
I don’t know: semantic or pragmatic marker?
Collins COBUILD English Language Dictionary (1987: 802) deªnes the verb know as follows: 1. If you know a fact, a piece of information, or an answer, you have it in your mind and are certain that it is correct; 2. … you have heard of something; 3. … you have information about a subject; 4. … you are familiar with a place, a work of art, an idea, someone, etc.; 5. … you are aware of something; 6. … you have the necessary skills and knowledge to do something. People say I know when they are familiar with a piece of information. When they use I don’t know, i.e. a declaration of insu¹cient knowledge, we need to distinguish a number of possible pragmatic functions. In her research, Tsui (1991) has identiªed a variety of functions of I don’t know. Here I limit myself to describing the ones that emerge from the data. She observes that I don’t know is generally taken to be a reply to an information question when the speaker is unable to supply the requested information: (5)
[…] why is the government going to sell more B T shares? I would have thought. <M02> I don’t know. <M01> with all these big proªts that’s money coming in for us isn’t it? (Cobuild: ukspok/04. Text: S0000000230)
However, Tsui has noted that I don’t know is produced even when the speaker is able to supply the information requested in the question, as in (7) and (8) below: (7) <M07>
[…] in anything else after twelve years of the sort of rule we haven’t isn’t it? <M01> Well I don’t know I mean I don’t think we can blame the government for all our ills I think it’s the way we are. (Cobuild: ukspok/04. Text: S0000000142)
161
162 Giuliana Diani
(8) <M01> <M04>
Don’t the two things sometimes go hand in hand? Well I don’t know. Well no I mean if some if one shop’s selling something for say a pound and another one’s selling […] (Cobuild: ukspok/04. Text: S0000000304)
In examples (5)–(7), the production of I don’t know is a way of avoiding that disagreement is expressed overtly and disclaiming what has been said in the previous turn. Prefacing a disagreement with a declaration of insu¹cient knowledge reduces the speaker’s commitment to the truth of the proposition expressed in the disagreement, hence mitigating its face-threatening eŸect. In general, disagreement sequences are structured so as to minimize the eŸects of explicit disagreement (Pomerantz 1984). An illustration is given in (8), where we see the addressee prefacing his disagreement with Well I don’t know, thus displaying reluctance and discomfort (cf. Tsui 1994). Throughout the examination of the data, I have noticed that I don’t know may serve as a ‘marker of uncertainty’, as Tsui (1991: 619) calls it. This is illustrated in (9) (= 2): (9) […] that can do a really good job like Maggie Thatcher. <M01> Now there seems to be I mean I don’t know this is not scientiªc at all but a bit of a sexual divide here. It seems the women like […] (Cobuild: ukspok/04. Text: S9000000810) By using there seems to be the speaker hedges the opinion that this is not scientiªc but a bit of sexual divide . I mean I don’t know further emphasises the speaker’s uncertainty and unwillingness to commit himself. Example (10) is somewhat diŸerent: (10) […] I mean how would we feel if erm I don’t know Holland owned the Isle of Wight or something like that I mean I suppose we’d feel a […] (Cobuild: ukspok/04. Text: S000000074) Here I don’t know seems to match Brown and Yule’s (1983: 109) description of ‘ªllers’, the principal function of which, they say, is ‘to ªll the silence and maintain the speaker’s right to speak, while he organizes what he wants to say’ (cf. Stenström 1984: 206). We can see from the data that I don’t know is also used for minimizing compliments. Examples (11) and (12) illustrate this: (11) <M01>
Well I’ll tell you something shall I Margaret? Yes please. Well speaking as a school governor
The discourse functions of I don’t know 163
Well I’m sure you qualify. Ooh I don’t know I mean er it frightens me the fact that we’re supposed to look after all this money and and run school and run it like a business. It worries me immensely you know. (Cobuild: ukspok/04. Text: S0000000156) (12) <M01> <M03>
You were lucky it’s worth a lot more than that now. Oh I don’t know er it’s it’s not something that I deal with because it was given to me by a friend […] (Cobuild: ukspok/04. Text: S0000000463)
According to Leech (1983: 136), there is a Modesty Maxim operating in conversation, which can be formulated as: ‘Minimize praise of self: maximize dispraise of self’. I don’t know, which serves as a polite de-intensiªcation of F01’s compliment, can be seen as an illustration of this maxim. Following Tsui’s analysis, we have seen that I don’t know does not necessarily occur in response to an information question. It can occur in a number of conversational environments, such as in reply to an assessment or a request, where it functions to: – – – –
avoid explicit disagreement; avoid commitment; minimize face-threatening acts; mark uncertainty.
In addition we need to investigate how certain ‘discourse markers’ (SchiŸrin 1987) occurring in conjunction with I don’t know aŸect its function. In the data, I don’t know is sometimes prefaced by well. In addition it occurs with other linguistic expressions such as oh, you know and I mean. These are discussed in the following section.
4.
I don’t know and discourse markers
SchiŸrin (1987: 31) deªnes discourse markers as ‘sequentially dependent elements which bracket units of talk’; i.e. they signal a relationship between immediately adjacent units of talk and therefore have a coherence-building function on a local coherence level. The following markers frequently co-occurred with I don’t know: well (194); oh (38); I mean (85); you know (23). Let us look at the markers in detail.
164 Giuliana Diani
4.1 Well5 Halliday and Hasan (1976: 269) observe that well “serves to indicate that what follows is in fact a response to what has preceded: in other words, it slips in quietly the respondent’s claim to be answering the question (…) and hence is purely cohesive in function”. SchiŸrin (1987: 102–103) argues instead that well “is a response marker which anchors its user in an interaction when an upcoming contribution is not fully consonant with prior coherence options”. It also allows a temporary suspension of other conversational expectations, especially the expectation of an immediately forthcoming answer, as in SchiŸrin’s example: (13) Sally: What does it mean:…if somebody has common sense. D’you have any idea Zelda: (We)ll:…I think if they have common sense um… [continues] (SchiŸrin 1987: 111) A similar view of well is expressed by R. LakoŸ (1973b: 461), who observes that well “is used in case the speaker senses some sort of insu¹ciency in his answer”. Well often occurs in disagreement sequences, thus “displaying reluctance or discomfort” (Pomerantz 1984: 72), or, according to GoŸman (1974: 546), “the response we often seek is not an answer to a question or a compliance with a request but an appreciation of a show put on”. This is illustrated in an example quoted from SchiŸrin (1987): (14) Henry: I don’t go out of my way eh because eh Zelda: Well you would go out of your way! (SchiŸrin 1987: 117) These functions of well are also illustrated in the corpus, where well prefaces I don’t know, signalling that the speaker does not give the expected or desired answer, as (15) and (16) show: (15)
[…] they’re they’re let out again and they do do it again. They’re
(16) <M01>
[…] Oh well old Jesus Christ would be happy to hear that wouldn’t he? Well I don’t know. Well I hope I still go up there to see him eventually
The discourse functions of I don’t know 165
<M01> Yeah
4.2 Oh Oh is traditionally viewed as an exclamation or interjection. When used alone, oh is said to indicate strong emotional states, e.g. surprise, fear, or pain (Collins COBUILD English Language Dictionary 1987). Examples (17) and (18) adapted from SchiŸrin (1987) illustrate this case: (17) Jack: Was that a serious picture? Freda: Oh! Gosh yes! (SchiŸrin 1987: 73) (18) Jack: Like I’d say, ‘What d’y’mean you don’t like classical music?’ ‘Oh! I can’t stand it! It’s draggy.’ (SchiŸrin 1987: 73) One of the basic goals of talk is the exchange of information. This goal can be realized in the communication situation by the redistribution of knowledge about entities, events, states or situations from the speaker to the hearer. Furthermore, because discourse involves the exchange of information, knowledge and metaknowledge are constantly in ¶ux, as are degrees of certainty about, and salience of, information. According to SchiŸrin (1987), oh marks a focus of speaker’s attention, which then also becomes a candidate for hearer’s attention. This creation of a joint focus of attention not only allows transitions in information management, but it marks information as more salient with a possible increase in speaker/hearer certainty as to shared knowledge and metaknowledge. A very similar view of oh is that of Heritage (1984: 299), who views oh as a particle “used to propose that its producer has undergone some kind of change in his or her locally current state of knowledge, information, orientation or awareness”. In the corpus there were 38 occurrences of oh combined with I don’t know. Let us consider extracts (19) and (20) below:
166 Giuliana Diani
(19)
Oh I I don’t know whether I agree a hundred per cent with that I mean you know you don’t know enough about diŸerent salaries and diŸerent wages but but certainly my experience has been that I have been as fairly paid in my profession as the man would be so I mean I don’t know the other side of of your argument […] (Cobuild: ukspok/04. Text: S9000000790)
(20) […] they’re competitive I mean they don’t cost any more than any other car park. <M01> Oh I don’t know. I mean where where I park in my local town very cheap ten pence an hour. (Cobuild: ukspok/04. Text: S0000000143) In (19) and (20) the information is acknowledged by means of the following structure: oh + I don’t know + … + I mean, that is, the orientation towards what has been said followed by a statement of insu¹cient knowledge and an attempt to present one’s opinion, which, from the very beginning, is bound not to coincide with the other speaker’s opinion. In (21), oh I don’t know has the eŸect of softening a forthcoming disagreement. The use of oh re¶ects the speaker’s realization at that point in the discourse that she does not agree with her interlocutor. I don’t know seems to function as an introduction to the expression of disagreement with M01’s opinion that having a heron in your garden is lovely. (21)
[…] some some use some useful use like like what? Oh I don’t know somebody that’s more needy than we than people that’s spending the money. <M01> Yeah (Cobuild: ukspok/04. Text: S9000000757)
In (22), M01 tries to get a clariªcation from M05: by using oh I don’t know M05 shows his unwillingness or inability to provide the information.
The discourse functions of I don’t know 167
Other pragmatic markers such as say, er, erm followed by I don’t know seem to function in the same way as the collocation oh I don’t know. When these expressions are used turn-internally, the speaker seems to recollect something, or to think of the best way to express something, as (23) shows: (23) […] It’s actually it’s quite nice like this but it’s not erm I don’t know. It would be comfy for me because I like sitting on things like this. I don’t think say like […] (Cobuild: ukspok/04. Text: S0000000465) In the next sections, I shall focus on two other markers, I mean and you know, which frequently occur in the corpus.
4.3 I mean In her study on discourse markers, SchiŸrin (1987) observes that the literal meaning of I mean and you know directly in¶uences their discourse use, which is conªrmed by this study. The corpus has 85 occurrences of I mean combined with I don’t know. Of these, 50 precede it and 35 follow it. There are several examples which support the hypothesis that such a marker increases the ‘tentativeness’ value of I don’t know, particularly when expressing personal evaluation: (24)
[…] people feel happy with you know dressing in the things that are the latest fashion I mean I don’t know I just feel happy wearing what I do. <M01> Yes
(25) <M01>
I mean yeah I don’t know how much did she make from the books?
In (24) I mean and I don’t know signal that the speaker is trying to ªnd a better way of expressing her assessment. In (25) I mean yeah I don’t know is followed by a whquestion. Example (26) is a case of self-correction:
168 Giuliana Diani
(26) <M01> I mean does it ¶avour the fruit at all?
In other words, I mean highlights the message which follows, which is strongly relevant to the overall interpretation.
4.4 You know You know is essentially a marker of information state. It implies that something is expected to be known both by the hearer and by the speaker. To this eŸect Östman (1981) points out that you know, like many other pragmatic devices, violates Grice’s (1975) Maxim of Quality, since “the lexico-semantic value of ‘you know’ would predict that what is said is in the knowledge of the addressee” (1981: 17). As a matter of fact, the speaker uses this marker in an attempt to have his interlocutor cooperate or to have him/her accept the propositional meaning of the utterance.
The discourse functions of I don’t know 169
The corpus contains 23 occurrences of you know combined with I don’t know. Of these, 16 precede it and 7 follow it. Here are some cases: (29)
Don’t forget sideways. So I don’t you know I don’t know what they’re going to do next. I don’t know what other little things she’s got up. (Cobuild: ukspok/04. Text: S9000000932)
(30) […] to open a factory in Eastern Europe. Now it’s only because it’s cheaper. <M09> I don’t know you know. <M01> And we’re exporting jobs aren’t we here. Be it in the coal mines or […] (Cobuild: ukspok/04. Text: S90000001105) In these examples, you know seems to function as an invitation to acknowledge a new piece of information.
5.
Conclusion
I have presented arguments to demonstrate that I don’t know realizes signiªcantly diŸerent pragmatic functions depending on the sequential environment. Despite the diŸerent pragmatic functions realized by the utterance, there is nevertheless a central meaning which uniªes all its occurrence: a declaration of insu¹cient knowledge. It is this semantic meaning that underlies its diŸerent pragmatic functions. Thus, in reply to an information question, a declaration of insu¹cient knowledge is a justiªcation for not supplying the required piece of information. Similarly, in reply to an assessment, a declaration of insu¹cient knowledge is an excuse for not making an assessment or for not agreeing with the speaker’s assessment, hence implying disagreement. A preface declaring insu¹cient knowledge, which introduces an act of disagreement or a negative assessment, reduces the speaker’s commitment to the truth of the proposition expressed, hence softening the face-threatening eŸect of the disagreement or the negative assessment. A factor which seemed important to investigate since it has not previously been investigated is how certain discourse markers occurring in conjunction with I don’t know aŸected its functions. The analysis has shown that: – –
well tends to preface I don’t know, signalling that the response will not be the expected or preferred one; oh, prefacing I don’t know, seems to display the speaker’s realization at a point in the discourse that the speaker does not agree with her/his interlocutor. Oh
170 Giuliana Diani
–
–
also tends to preface I don’t know when the speaker is hesitating while talking, for example because s/he is not sure of the facts or is searching for the right word; I mean, prefacing I don’t know, increases its ‘tentativeness’ value, particularly when expressing personal evaluation; when following it, I mean seems to modify the speaker’s initial declaration of insu¹cient knowledge so as to support her/his argument; in collocation with I don’t know, you know seems to be used as an invitation to acknowledge a new piece of information.
The present study does not claim to be exhaustive, not only because it does not exclude that the use of I don’t know is also pertinent in other conversational exchanges that the corpus has not highlighted, but also because it is di¹cult to provide clear-cut distinctions between diŸerent functions.
Notes * I wish to express my sincere thanks to Karin Aijmer, Anna-Brita Stenström, Marina Bondi and Louann Haarman for many valuable comments on earlier versions of the paper. 1. For a detailed study of the principles for constructing polite speech, see Brown and Levinson (1987). 2. For further information on the corpus, see http://www.titania.cobuild.collins.co.uk 3. Transcription conventions: <Mn>
References Brown, Penelope and Levinson, Stephen C. 1987 Politeness: Some Universals in Language Usage. Cambridge: Cambridge University Press.
The discourse functions of I don’t know
Brown, Gillian and Yule, George 1983 Teaching the Spoken Language. Cambridge: Cambridge University Press. Collins COBUILD English Language Dictionary 1987 Edited by John Sinclair et al. London, Glasgow: Collins. GoŸman, Erving 1967 Interaction Ritual: Essays on Face-to-face Behaviour. New York: Doubleday. 1974 Frame Analysis. New York: Harper and Row. Grice, Paul H. 1975 “Logic and conversation”. In Syntax and Semantics 3. Speech Acts, P. Cole and J. L. Morgan (eds), 41–58. New York: Academic Press. Halliday, M. A. K. and Hasan, Ruqaiya 1976 Cohesion in English. London: Longman. Heritage, John C. 1984 “A change-of-state-token and aspects of its sequential placement”. In Structures of Social Action: Studies in Conversation Analysis, M. J. Atkinson and J. Heritage (eds), 229–345. Cambridge: Cambridge University Press. LakoŸ, Robin 1973a “The logic of politeness; or, minding your p’s and q’s”. In Papers from the Ninth Regional Meeting of the Chicago Linguistic Society. University of Chicago, 292–305. 1973b “Questionable answers and answerable questions”. In Issues in Linguistics: Papers in Honor of Henry and Renée Kahane, B. B. Kachru et al. (eds), 453– 467. Urbana: University of Illinois. Leech, GeoŸrey 1983 Principles of Pragmatics. London: Longman. Östman, Jan-Ola 1981 You know: A Discourse-Functional Approach. Amsterdam: John Benjamins. Pomerantz, Anita 1984 “Agreeing and disagreeing with assessments: some features of preferred/ dispreferred turn shapes”. In Structures of Social Action: Studies in Conversation Analysis, M. J. Atkinson and J. Heritage (eds), 57–102. Cambridge: Cambridge University Press. SchiŸrin, Deborah 1987 Discourse Markers. Cambridge: Cambridge University Press. Stenström, Anna-Brita 1984 Questions and Responses in English Conversation. Malmö: Liber Förlag. Tsui, Amy B. M. 1991 “The pragmatic functions of I don’t know”. Text 11 (4), 607–622. 1994 English Conversation. Oxford: Oxford University Press. Widdowson, Henry G. 1983 Learning Purpose and Language Use. Oxford: Oxford University Press.
171
“They’re a little bit diŸerent”… Observations on hedges in academic talk Anna Mauranen University of Tampere
1.
Introduction
Research into hedging was initially stimulated by George LakoŸ’s (1973) seminal paper which deªned hedges as expressions which make propositions more or less fuzzy. He dealt with the semantic and logical consequences of hedge use on propositions. Other early research focused on pragmatic meanings in spoken interaction (Prince et al. 1982; Hübler 1983). Gradually the main focus of hedging research has shifted from the initial logico-semantic origins towards the pragmatic aspects, although traces of both approaches can be seen in most studies — diŸerent deªnitions generally incorporate elements of both inexactitude and mitigation. In the 1990s the ªeld became dominated by politeness theory, and generally a somewhat socio-psychological approach. At the same time, interest also turned from hedging in conversation towards hedging in academic research papers (see, e.g. Hyland 1998; Markkanen & Schröder 1997). Despite these changing emphases in both data and theoretical orientation, most scholars in the ªeld have sought to ªnd the common core of all hedges, with a broad agreement that the expressions are usually multifunctional. In this paper, I wish to take an alternative approach and explore variation in hedges with a basically micro-level approach. What I am trying to do is to take some typical hedges and explore their use proªles in the Michigan Corpus of Academic Spoken English (MICASE; see, Simpson et al. 1999). My point of departure is that there is a basic distinction between what I have called the ‘epistemic’ and ‘strategic’ (or ‘interactive’) functions (Mauranen 1997; for a similar distinction see also Hyland 1998). I explore individual expressions and see to what extent the distinction is relevant in their usage, and to what extent their primary use falls into one or the other category.
174 Anna Mauranen
The as yet smallish amount of the data and the nature of the questions asked render this an exploratory survey. The basic question is how far it makes sense to distinguish between ‘epistemic’ and ‘strategic’ uses, and to what extent we can speak of individual proªles of hedging expressions, in terms of this distinction or other criteria emerging from the study. Although the context in the MICASE data is academic, the setting is that of university discourses in general rather than exclusively the research community. For comparison, speech data from other spoken corpora, mainly a million word sample of the spoken part of the BNC (included in the BNC Sampler, 1999), and also in some cases the spoken part of the Bank of English, are consulted. The analysis focuses on expressions which commonly appear in the hedging literature. These fall into two main types: ªrst, expressions of vagueness, including ‘vague category tags’ (Channell 1994), such as or something and or so, and other markers of imprecision, such as sort of, kind of; and secondly, mitigators such as just and a little bit. The ªndings suggest that (1) diŸerent hedging expressions have their speciªc proªles of use, (2) many of these expressions have diŸerent functions depending on whether they occur in a strategic or an epistemic context. Thus, it appears that for instance expressions like a little bit or just are primarily used for strategic purposes, i.e. for softening and mitigation, while kind of, sort of and or something have more epistemic uses, indicating conceptual openness rather than redressing a threat to face. In strategic uses many of these appear to be oŸsetting a negative implication of some kind. It is also interesting to note that some of the expressions are used pseudo-epistemically or pseudo-strategically, that is, the instances fulªl the criteria for one category, but the context induces the other interpretation as well.
2.
Background
Hedging is generally taken to mean those expressions in language which make messages indeterminate, that is, convey inexactitude, or in one way or another mitigate or reduce the strength of the assertions that speakers or writers make. Typical expressions (perhaps, sort of, just, seem, etc) make assertions less categorical and direct, and are thereby regarded as less certain, as well as usually more polite than would be their unmodiªed, direct counterparts. The ªrst opening into hedging by George LakoŸ (1973) was basically semantic and logical, dealing with propositions and their degrees of fuzziness. Very soon interest arose also in hedging as a speech strategy (R. LakoŸ 1975, Prince et al. 1982), which represents a pragmatic viewpoint, and has drawn scholars with interests in pragmatics and speaking.
“They’re a little bit diŸerent”…
Scholarly interest in hedges has re¶ected the dual origin of the concept through the two or three decades that hedging has been on the research scene. Although the division between an epistemic and a strategic interpretation of hedging (or modality as an overlapping category) has been present all along, some linguists prefer to look for unity rather than division between them. For example Halliday (1994) treats modality as part of the interpersonal function of language, and so does Simpson (1990; 1993). Simpson deªnes modality as the speaker’s attitude towards the truth of a proposition expressed by a sentence and the situation or event described in that sentence. The deªnition is clearly epistemic, but this epistemic interpretation then gets connected to politeness: “In addition to the speaker’s uncertainty, modalized utterances are motivated pragmatically by tentativeness, tact and distance — in short, considerations of politeness” (Simpson 1990: 73). In the 1990’s, a good deal of attention shifted to hedges in written academic discourse (e.g. Myers 1989; Salager-Meyer 1994; Markkanen & Schröder 1997; Crompton 1997, 1998; Hyland 1998). This research is characterised by an integration of the propositional role of hedges as modifying the precision or certainty of statements with the interpersonal function of politeness, in the lines of Simpson quoted above. Thus, while the role of hedges as increasing or decreasing conceptual fuzziness has been recognised, at the same time it has been seen as subservient to the signalling of interpersonal relations, above all face-saving and politeness (see for example Myers 1989; Markkanen and Schröder 1989; Simpson 1990; Varttala 1999, 2000). These analyses draw on Brown and Levinson’s notions of politeness and face, some of them focusing on expressions of modality (e.g. Myers 1989; Simpson 1993), which in these studies ªt in with the domain of hedging. Yet the two views do not seem easily compatible. The reason is that the need to express matters of precision and imprecision in natural language is important and manifold in itself (see, e.g. Channell 1994) and go beyond the needs of managing interpersonal aŸairs. Although it clearly overlaps with the need to avoid face loss for either speakers or listeners (Markkanen & Schröder 1989, 1997), or the need to increase the acceptability of propositions (Meyer 1997), instances remain where a strategic motivation for hedging appears far-fetched. Academic writing provides many examples of this — there are many things that a scientist or scholar does not or cannot know; for instance, even if we are able to isolate some plausible causal relationships in the humanities or social sciences most scholars, following currently accepted epistemological beliefs, would be reluctant to say that these exhaust all possible in¶uencing factors. There is also a need to mark out a statement as a hypothesis, or speculation. This is not necessarily evasive or apologetic, implying that the author tries to avoid saying anything that might later turn out to be false. There is simply a need to indicate hypotheticality as important information about
175
176 Anna Mauranen
the status of a statement. Moreover, ideologies of scientiªc and scholarly enquiry do not imply that being shown to have been wrong necessarily means a loss of face — on the contrary, views and results being replaced by others is a normal part of the game. The usual conceptualisation of academic writing as a battleªeld, although clearly part of the truth, is insu¹cient: a collective eŸort to make sense of the world is intertwined with the competitive facet. Scholars and scientists need to be able to see and indicate what they regard as established facts and what they regard as controversial, or epistemically open, as it were. I would therefore like to maintain the distinction between ‘epistemic’ and ‘strategic’ (or ‘interactive’) motivations for hedging, as I have argued before (Mauranen 1997). Even though all instances may not in practice be comfortably assigned to one category without evoking a strong possibility of interpretation in terms of the other, the fact that expressions can be, and often are, multifunctional does not invalidate the postulation of diŸerent functions in itself. The distinction between primarily epistemic and primarily strategic uses is therefore applied in this study, as a preliminary tool for dealing with the uses of some hedges in academic speech. A primarily epistemic hedge is one which aŸects the propositional content of a statement: (1) but now with this factor, the, m- ma- the scope is maximized, at a somewhat, lower, temperature A primarily strategic hedge is one where the propositional content is either irrelevant or hard to detect: (2) perhaps er we’ll we’ll pursue this concept during the next er hour or so. This distinction is not unlike that put forward by Hyland (1998), who distinguishes between content-based and interaction-based hedges. A couple of speciªcations to the primary distinction between epistemic and strategic hedging need to be made. First, although the propositional eŸect is the starting point for the epistemic use, it is clearly too narrow. Expressions which fall into the domain of language use which could be called ‘ideational’ following Halliday (e.g. 1994) or the ‘autonomous plane’ in Sinclair’s (1983) terms, i.e. which deal with the referential function of language rather than the interactive or strategic function, constitute a larger category than that which can be assessed by reference to truth value. For example, it is not useful to categorise evaluative descriptions (as in 3, describing a work of art) as strategic if the relations of the participants in the discourse situation are not involved. This is treated as epistemic, then. (3) it’s just so, kind of earthy and real and and of the essence of, of life.
“They’re a little bit diŸerent”… 177
The epistemic category therefore comprises primarily referential uses in addition to propositional ones. Secondly, the problematic ‘third’ in accounts of language, the ‘textual’ function, needs to be placed somewhere, since hedges seem to prefer the company of discourse re¶exive (or metadiscursive) expressions. Since such ‘collocative’ behaviour of two functional categories has looked like mitigating the imposition of the speaker’s order on the ongoing discourse (see e.g. Mauranen 2001; 2002), I have followed this interpretation and taken hedges accompanying re¶exive discourse (as in 4) to be strategic. (4) when we get to the second half of the course, i’ll be uh sort of winging it even more than i usually am. As I already said, this functional dichotomy does not imply a strict division with each instance assignable to one and only one category — the basic functions have an overlap area where instances can be used pseudo-strategically or pseudoepistemically (example 5). In other words, they ªt the criteria of one category but are used in the discourse in the other way. Clearly, this provides a good deal of ¶exibility and possibilities for expressing a wide range of meanings with a small selection of items. (5) that’s that’s, sort of like, this is sort of more of a, trade oŸ or cost beneªt, (mhm) you know In (5), the speaker is giving an ad hoc descriptive label to the concept he is pursuing. Thus the hedge would qualify for an epistemic interpretation, but the repeated hedging and hesitation, together with the appeal to hearers (you know) makes it sound as if the emphatic uncertainty was intended to reduce the speaker’s commitment to what he is saying. The hedging may also make the utterance interactively more appealing by leaving the ¶oor open for other participants to accept or reject this interpretation. In the context of a meeting, where the utterance was made, it seemed in eŸect to make a suggestion that the situation be interpreted in these terms. It was therefore regarded as pseudo-epistemic, making use of the two functions simultaneously. The example shows some of the di¹culty involved in trying to impose categorisation on complex and multifunctional phenomena, even with allowing overlap and gradual shading of the functions into one another. The point of even making the eŸort lies in not accepting unnecessary generalisations concerning a generic integration of the epistemic and strategic functions. To see the extent of the integration, it is necessary to try to maintain their separation as far as possible. What I am trying to do in this paper is to take some typical hedges and explore their use proªles in the MICASE corpus. To investigate two basic aspects of hedging,
178 Anna Mauranen
I have studied ªrst of all some hedges which characteristically convey vagueness or inexactitude, such as sort of /kind of, and or something, or so. The latter are called ‘vague category tags ‘(Channell 1994) or ‘coordination tags’ (Biber et al. 1999), and as Biber et al. (1999: 115) put it (hedgingly): “are best described as some kind of vagueness markers or hedges”. For the second group of hedges I chose some that seem to have the potential for serving mitigating functions (a little bit, somewhat, just). With the choice of these two types I hope to capture something of the variance of individual hedging expressions, and to explore the usefulness of the epistemic vs. strategic distinction in analysing hedging.
3.
Data and methods
The MICASE corpus (Simpson et al. 1999) consists of a variety of speech genres recorded at the University of Michigan (for a detailed description, see http:// www.Isa.umich.edu/eli/micase/micase.htm). The selection criteria have emphasised wide coverage of the speech used in academic contexts, thus excluding only speech events that are not relevant to the institutional role of the university, such as service encounters in cafeterias, etc. Field trips, tutorials, colloquia, discussion groups, and similar events were included along with the more ‘core’ activities like lectures, seminars, and thesis defences. The present data represents the corpus at the stage when half a million words were transcribed (out of the ªnal target of 1.5 million), and therefore the ªndings may be less reliable than those made on the basis of the ªnal corpus, for example by being in¶uenced by more speaker or topic bias. Nevertheless, spoken corpora generally tend to remain much smaller than written corpora, and of course merely increasing the size will not solve problems of representativeness. A small corpus size is not as detrimental to a specialised database as to a general corpus, as its claims to representativeness are more focused. The hedging expressions to be studied were chosen from among those commonly discussed in the literature and presented as typical hedges in two categories representing the main functions of hedges: imprecision markers (kind of, sort of, or something, or so) and mitigators (somewhat, just, a little bit). Although modal expressions are also frequently quoted as typical representatives, they were left out of this study because they have received much more attention from various research perspectives, and because the less conspicuous yet typical hedges discussed here tend to get lumped together and receive less diŸerentiated treatment than modals. In most cases, all occurrences of the expressions could be taken into account, but with the most frequent items, sampling was resorted to. The data was split into ‘lectures’ and ‘other genres’, because at this stage of the corpus, lectures were the clearest and largest monologic speech event type. Lectures also have a
“They’re a little bit diŸerent”… 179
speciªc task of conveying information, which sets up the expectation that there is a greater need for epistemic hedging than in more overtly interactive and dialogic event types. The lectures database was smaller, and thus the expectation is that each expression is used less frequently in lectures than in ‘others’. As I already explained in the preceding section, the epistemic function was taken to include hedging relevant to the referential aspect of utterances. In strategic uses I have included interactive suggestions, clariªcations, interpretations, and similar acts which involve a direct negotiation between interlocutors. The utterances characteristically have deictic expressions, mostly ªrst and second person pronouns. For every expression, I have compared the MICASE ªndings to the spoken section of the BNC Sampler (1999). I am thus assuming that the BNC Sampler acts as a kind of standard, as it consists of general speech data, and it is only about twice as large as the version of MICASE discussed here. The expected numbers of occurrences in MICASE would then be about half the number in the Sampler. Clearly the comparison is only very rough, given that both corpora are smallish, and that there are two important diŸerences between them: the genre selection, which is focused narrowly in MICASE and broadly in the BNC, and the variety diŸerence between British and American English. It is therefore not possible to explain diŸerences in ªndings in respect of either of these divisions. However, I wanted to make the comparisons in order to detect possible major diŸerences, which could then form the basis for hypotheses for later research looking into either academic and other kinds of speech or British and American varieties.
4.
Vagueness indicators
Kind of and sort of are nearly invariably included in typical hedges indicating fuzziness, imprecision, approximation etc. They are also usually treated together, as variants of basically the same meaning. I have here dealt with them separately, but they did not show any noticeable diŸerences, apart from some distributional dissimilarities, sort of being more common. This diŸerence was consistent across the genre division (lectures and other genres), and although there was one speaker who exceptionally favoured the expression, she cannot distort but one ªle. However, this cannot be considered important on the basis of the present evidence alone.
4.1 Kind of To begin with kind of, which, not unexpectedly, occurred frequently in lectures but even more often in other genres, it was predominantly used in the epistemic function (for a quantitative summary, see Table 1). In declarative utterances, it
180 Anna Mauranen
mainly preceded an ad hoc descriptive label — sometimes categorising (6), but mostly just descriptive (7–9). The eŸect is to leave the category epistemically open. The impression in (6) is that the work of art considered is somehow not a typical triptych, or that there might be other characterisations that ªt the piece equally well. The other examples seem to be oŸering a characterisation of the referent, one that will serve the purpose for the moment but is not to be taken as conclusive. (6) here’s a, kind of a triptych, by Joan Mitchell (7) and perinatal mortality statistics, they serve as kind of a gloss for how advanced a civilization is (8) you know it’s just so, kind of earthy and real and and of the essence of, of life. (9) what the dative does is put a kind of, reference or speciªcation, for these In strategic uses, like in suggestions (10) or in discourse re¶exive contexts, kind of seems more of a minimiser or mitigator (11), but even with re¶exive discourse it can be used in the same way as above: preceding an ad hoc descriptive label (12, 13), which in these cases leaves open exactly the kind of speech act that the speaker has in mind, merely giving it an approximate characterisation. At the same time, the speaker’s intention, and his or her organising in¶uence on the discourse situation is rendered tentative — thus leaving the interaction more open than it would have been without the hedging. This strategic openness is thus parallel to the epistemic openness. (10) that’d be kind of an interesting thing for you to, study if you’re talking about gender, relations (11) let me just kind of refresh your memory about the word abstract. (12) the ªrst thing that, i wanna do is kind of, go through and talk about these general stages (13) i’m gonna say more about this later on i just wanna, kinda do an overview so… A methodological comment I would like to make here relates to evaluative adjectives: unless they appeared in a dialogic context relating directly to the interactants and their mutual relations, I treated them as epistemic, as for example (14) and (15): (14) and then that Coolio song was kinda annoying (15) these are pretty low numbers so it’s kind of weird to see it in this form I chose to see these as epistemic not because they really have truth value, but because they deal with ‘content matter’, i.e. the referential content of the utterance
“They’re a little bit diŸerent”…
rather than the interactive potential. The analysis is also in line with those evaluative adjectives that are followed by nouns. Evaluation, despite certain ‘subjectivity’ involved, is then not in itself taken to imply such emphasis on interactivity as to render the expression strategic. Comparing MICASE with BNC speech, the diŸerence was striking. Proportionally, MICASE data had four times as many occurrences of kind of, and in addition, most of these were hedges, whereas the normal (4/5 of the examples) use of this sequence in the BNC data was not hedging, but of the structure X kind of Y. A random sample of the ten million-word spoken part of the Bank of English was similar to the BNC: only 16 instances of 50 were hedges. Since it is hard to imagine that hedging with kind of would be typically academic as compared to spoken language more generally, the more likely explanation is American usage as opposed to British. Table 1. Summary of kind of epistemic
strategic
other
total
79 (67) 60 (30) 18 (9)
14 (12) 8 (4) 2 (1)
7 (6) 32 (16) 80 (40)
100 (85) 100 (50) 100 (50)
MICASE lectures % (total of 85) other genres % (sample of 50/340) BNC sampler spoken % (sample of 50/212)
In this table, as in the others, percentages are presented along with the absolute ªgures to facilitate comparability, although the numbers are very small. The numbers are not directly comparable, since some are from totals varying in size, others from samples.
4.2 Sort of Moving on to sort of, it was on the whole very similar to kind of. The epistemic uses are the most typical (16–19), providing a descriptive label or illustrative expression, thereby emphasising the descriptive eŸort or quality involved. (16) okay? my sort of, semisolid deªnition of life, would be… (17) you can hear him trying to sort of translate what the epidemiologist told him… (18) they do it in a in a very sort of dispassionate way… (19) that would be an illusion in the, in the sort of objective reference, sense.
181
182 Anna Mauranen
In the lectures, all strategic uses were discourse re¶exive (20, 21), but in other genres not necessarily (22): (20) so i’ll put a little arrow here, just sort of, saying those those semantics again. (21) okay let me get into sort of the more serious stuŸ (22) and, i think sorta just leave leave it the way it is, okay. The BNC did not diŸer much from the MICASE corpus (see Table 2). Instead, two clear diŸerences emerged in relation to kind of in the BNC itself. First, there were four times as many cases of sort of as kind of, and secondly, over half of the sampled instances were used as hedges. In line with the ªndings above, the use was predominantly epistemic (as in examples (23–25). (23) they all just sort of sat there (24) so it was sort of going to the corner shop and buying whatever they could (25) I sort of bounced about, she sort of bumped along The strategic cases were suggestions: (26) well perhaps we could do it in a sort of ¶oor by ¶oor actually (27) could you please sort of pass it fairly quickly It seems that the hedging function may be more characteristic of sort of than kind of, although in American speech both appear to be used a good deal. As to actual meanings in hedge use, sort of and kind of seem reasonably similar. Both are largely used preceding ad hoc labels or descriptions or evaluations, and seem to presuppose that the hearer does not know what the speaker means without an eŸort on the speaker’s part to describe it; i.e. the information in the description is ‘new’ not ‘given’. This interpretation of kind of and sort of can be likened to Poos & Simpson’s (1999) analysis, also based on the MICASE corpus, where their ªndings included examples of kind of and sort of frequently preceding the use of particularly sophisticated jargon words (kind of a Faustian bargain) as well as metaphors. Poos & Table 2. Summary of sort of epistemic
strategic
other
total
56 (28) 50 (25) 48 (24)
18 (9) 14 (7) 8 (4)
26 (13) 36 (18) 44 (22)
100 (50) 100 (50) 100 (50)
MICASE lectures % (sample of 50/131) other genres % (sample of 50/424) BNC sampler spoken % (sample of 50/928)
“They’re a little bit diŸerent”… 183
Simpson suggest the jargon use might be explained as a gesture of solidarity with less knowledgeable hearers; with the metaphor use their interpretation is that speakers use hedges as metapragmatic markers drawing the listener’s attention to the non-literal terminology. To me, all of their examples in these categories would nicely ªt into the ad hoc signalling function of kind of and sort of.
4.3 Or something Or something is less prototypical as a hedge than kind of/sort of, but as a ‘coordination tag’ (Biber et al. 1999) or ‘vague category tag’ (Channell 1994) it is, along with or so, used for indicating vagueness, much like sort of and kind of, even though the syntactic possibilities are dissimilar. I have earlier argued (Mauranen 1997) that vagueness is part of epistemic hedging, but this statement needs to be modiªed in the light of the present data. As we already saw with sort of and kind of, epistemic meanings dominate (cf. Table 3), but do not exhaust the uses of vagueness indicators. I have included in the analysis here instances with like that following, although the form or something on its own was far more common, so the whole unit considered is or something (like that). This tag can be appended to a nominal head (28–30), a clause, or a numeral. (28) like, it dissolves in, if you add hydrochloric acid or something (29) just pick some weird herb and put that in there too, some tansy or something like that. (30) in traditional painting, you would have the warm, colors the haystack or the cottage or something, i- i- in the front of the painting … Mainly it appears to indicate that an instance is being talked about as a representative of a type or a category which is not speciªed, but rather exempliªed, as in the instances above. The category may be self-evident or otherwise di¹cult or unnecessary to specify, in general not important for the purposes of the present communication. This function could be called ‘unspeciªed category marking’ in that the instances imply a category by exemplifying but not specifying. When the tags were attached to a clause, it was typically a reporting clause (31–33), although not invariably (34). The reporting clause was practically every time reporting what might be termed imaginary discourse — a generalisation, or a hypothetical utterance. In this way its exemplifying character was similar to the nominal tag: both deal with category illustration without specifying the category itself. (31) you may say the question doesn’t make sense or doesn’t arrive or something but, but
184 Anna Mauranen
(32) she’d hand you some authors and say go ªnd your books or something (33) i promised not to is to say hah you’ve conceded my entire world view or something (34) it’s almost everybody until you get sensible (and) turn thirty or something Although quite obviously the usual place of the unspeciªc category indicator was after its head, an exception was also found — perhaps typical of the incremental nature of adding units in speech instead of working with ‘logical’ constituents: (35) if, they wanted to, apply or something like that for funding then it’d be a, good idea to Occasionally or something indicated approximate number or date, giving the order of magnitude (forty years or something; in the tenth grade or something; point-ªve point-six point-seven or something like that). Although strategic uses were very rare, they did occur (36, 37): (36) meet that day or something? (37) do you have a pen or something? (37) shows the di¹culty of keeping the epistemic and strategic functions strictly apart: the question is clearly strategic, but the use of the tag, if taken by itself, is exactly like the unspeciªc category markers above. The category that ‘pen’ belongs to is not presented, but presumed, and ‘pen’ exempliªes it. Here the classiªcation as strategic was made on the basis of the speech act it was part of, but a diŸerent solution could have been motivated. There are areas where the uses merge, even if the basic or prototypical cases of each category are not confusable. The BNC examples were also mainly epistemic. Where they departed from MICASE was in the greater frequency of ‘event’ as opposed to ‘nominal’ types, i.e. the tags were attached to clauses (38–40) as often as to nominal heads. (38) That ‘s right, they’ve probably had a row half way there or something, (39) I think it was Mrs said oh good have you come to take them away or something like that (40) they can’t determine it or something until he ‘s a few months The nominal type was indicating a category by exemplifying in these everyday examples just as in the academic cases: (41) probably a, a a paperback horror or something . (42) they ‘ll have sandwiches or something for tea (43) I do n’t know he ‘s some ¶u or something The number references again gave an estimate of the order of magnitude (more like three quid or something).
“They’re a little bit diŸerent”… 185
Table 3. Summary of or something epistemic
strategic
other
total
82 (14) 80 (92) 84 (42)
– (–) 17 (20) 12 (6)
18 (3) 3 (3) 4 (2)
100 (17) 100 (115) 100 (50)
MICASE lectures % (total of 17) other genres % (total of 115) BNC sampler spoken % (sample of 50/231)
4.4 Or so In principle, or so appears to be very similar to or something. In fact, it is much less frequent and much more restricted, so that in MICASE it was entirely limited to qualifying numerals, expressing approximate quantity or approximate time. What unites or so and or something is the ‘approximator’ role, and both can be used interchangeably in some numerical and time reference contexts. Or so was most often used with approximate time (44–46). (44) i have everything pretty much wrapped up in here by Wednesday or so (45) like within the last ten ªfteen years or so, (46) he’s- s- studied the MAR for i don’t know, ªfteen or so years, The other large category was approximate number: (47) V-perpendicular T is on the order of, ten to the minus two or so, (48) was something of the order of three hundred or four hundred or so angstroms thick or thin… (49) so like in this room there’d be about, what ªve times six or so thirty thirty or so modalities… The British speech examples from the BNC represented the same categories of use as those above, that is approximate time (50–52) and approximate number (53– 54), the ªrst being more common. (50) (51) (52) (53) (54)
ªrst week or so we ‘ll have to live in that back room lovely and cosy after half an hour or so in the last six months or so It ‘s about thirty pence or so usually it ‘s about a dozen or so people meeting once a month
What was diŸerent, however, was that there were also some strategic uses, like (55):
186 Anna Mauranen
(55) perhaps er we’ll we’ll pursue this concept during the next er hour or so. On the whole, or so and or something share an ‘approximative’ function. However, or so is almost entirely limited to numerical expressions, while or something is not. Or something implies a category by exemplifying and therefore acts as an unspeciªc category marker. The category is supposed to be retrievable by the hearer. In contrast, kind of and sort of seem to imply that the hearer will not be able to retrieve what is in the hearer’s mind without speciªcation or description. They precede much more frequently a description than a category label or instance. So they can be characterised as preceding an oŸer of new rather than given information — or, presenting rather than presuming reference. All the imprecision and vagueness markers are used primarily in an epistemic function; although epistemic meanings dominate, they do not exhaust the uses of vagueness indicators. In strategic use they seem to have a mitigating eŸect, that is, to oŸset a threat or to balance out some other negative potential arising from the situation. In their epistemic uses, however, no connection to negativity is discernible. Table 4. Summary of or so epistemic
strategic
other
total
100 (3) 82 (18) 78 (31)
– (–) – (–) 18 (7)
– (–) 18 (4) 5 (2)
100 (3) 100 (22) 100 (40)
MICASE lectures % (total of 3) other genres % (total of 22) BNC sampler spoken % (total of 40)
5.
Mitigators
The second main group that I have explored in this paper consists of hedges which do not so much seem to convey imprecision as to reduce the strength of the utterance. These expressions are therefore better candidates for the strategic function, since it is easier to interpret the need to use such expressions as interactively motivated. As before, I chose some expressions which are generally considered to be characteristic hedges from the type that appears to have a downtoning function. The expressions resulting from this somewhat random procedure are somewhat, a little bit, and just.
“They’re a little bit diŸerent”… 187
5.1 Somewhat To begin with somewhat, it is not a very frequent item in any of the data analysed. Yet it is relatively more common in MICASE: the total of occurrences was 23, which compared to the BNC’s 17 is about three times proportionally as frequent (cf. Table 5). It was used epistemically more often than strategically, which goes against the initial assumption that downtoners would be strategic. Examples from lectures would be (56) to (58): (56) so the two disciplines, seem to share very little granted these are somewhat stereotypic uh notions or deªnitions (57) that’s a, a somewhat arcane limiting factor but but it illustrates the point. (58) this presents a a general and much ki- and somewhat controversial issue of which are the most dominant features. In other than lecture genres, the uses appeared also predominantly epistemic (59–61). (59) you drew direct conclusions from them, but then to extrapolate them to general conclusions about culture and cognition and science, is, somewhat conjectural… (60) although this_ this may seem like a somewhat limiting um, um, set of, utility functions it turns out… (61) uh, it is it is s- s- somewhat of an accumulated failure, uh because you see… The contexts in which somewhat was used were clearly negatively biased: ªve out of eight instances in lectures, and nine out of 15 in other genres were negative, and the rest not positive but either concrete or otherwise fairly neutral. A comparison with the BNC data shows a very similar usage, although there are fewer cases (examples 62 to 64): (62) our membership slid somewhat and we haven’t got a very big membership in er Water now (63) the conditions on the ground ¶oor are somewhat diŸerent. (64) within the next twelve months if not somewhat earlier now. The uses were again predominantly negative, (10/17), but also neutral, in concrete contexts particularly, as in (64) above. The only recurrent pattern in the data was somewhat diŸerent. In sum, somewhat thus seems to act indexically as a forewarning of a possibly negative evaluation, and be epistemic most of the time. A negative prosody (see Louw 1993) with somewhat was also discovered by Stubbs (1996), with quite diŸerent data.
188 Anna Mauranen
Table 5. Summary of somewhat in epistemic and strategic functions epistemic
strategic
other
total
75 (6) 87 (13) 76 (13)
13 (1) 13 (2) 24 (4)
13 (1) – (–) – (–)
101 (8) 100 (15) 100 (17)
MICASE lectures % (total of 8) other genres % (total of 15) BNC sampler spoken % (total of 17)
Table 6. Summary of somewhat in terms of positive and negative connotations positive
neutral
negative
unclear
total
– (–) – (–) – (–)
38 (3) – (–) 35 (6)
63 (5) 60 (9) 47 (8)
– (–) 40 (6) 18 (3)
101 (8) 100 (15) 100 (17)
MICASE lectures % (total of 8) other genres % (total of 15) BNC sampler spoken % (total of 17)
5.2 Just The next item in the mitigating function was just, which unlike somewhat was extremely frequent. To choose just as a typical hedge is against its description in dictionary entries and grammars, but its high frequency attracted attention at the very early stages of MICASE compilation, and on inspection the pragmatic functions, above all the minimising one, emerged immediately. For the sake of consistency, I shall here limit myself to an analysis similar to those of the other hedges, but a more ªne-grained break down into more subtle uses, supplemented with case studies and a correlation with pronunciation patterns is found in Lindemann & Mauranen (2001). Just was somewhat more common in the MICASE than in the BNC data, but the diŸerence was only about 1.6 times in favour of MICASE, and given the uncertainties involved, this is not strong evidence of an important diŸerence. Since there were large numbers of occurrences (2,782 in MICASE and 3,568 in the BNC), I resorted to sampling from both corpora. The sampling solution may not be entirely happy in that it overrepresents lectures from MICASE: I sampled 50 from both lectures and other genres, although there was less data in the lectures. However, since I wanted to explore the possible diŸerences in use between these two domains as I have done with the other items, and since I am not basing my argument on small diŸerences in the data (which is not yet reliable for such
“They’re a little bit diŸerent”… 189
analysis), it will not seriously distort the results. Nevertheless, it needs to be kept in mind. The epistemic uses of just were typically ‘minimisers’, i.e. they limit the scope or application of a concept (65–66) or reduce its signiªcance (67). This was also the most frequent category in the use of just. Other uses included ‘particularisers’ (Lindemann & Mauranen 2001), which could roughly be glossed ‘exactly’ as in (68). These might perhaps be seen as re¶ecting the role of hedges as making things more precise, as G. LakoŸ (1975) suggested, but it is more natural not to include these instances in hedges at all, and the present deªnitions of inexactitude and mitigation certainly exclude them. The remaining two categories of just, ‘emphasisers’ and temporal meanings were naturally not included in hedges. In all, the hedging uses of just were less dominant than in the case of somewhat, which did not have instances of other uses, but still a clear majority of about 4/5 were hedges (38/50 in lectures, 40/ 50 in other genres). (65) but um, i’m wondering if they want you to take out of that alcohol-related that’s in some other category too, or if it’s just supposed to be sort of aggregate, alcohol-related. (66) to appreciate it’s just all, one dimensional it’s all coming at you that way visually you’re not, (67) and even leads to the whole problem of, intermediate stages or whether it was just basically a leap, from the front to the back. (68) what do we ªnd?… that’s just, one over I H-bar, times, H-bar omega J, alpha J Just appeared more often in epistemic than strategic meanings in the lectures, but about equally in the non-lecture genres. The strategic hedges in lectures tended to co-occur with discourse re¶exive, or metadiscursive expressions, thus mitigating the speaker’s imposition of his or her order on the discourse (69 and 70). In other speech events the strategic uses were toning down criticism (71, from a thesis defence) or softening suggestions (72, from a meeting). (69) by the way have been some in the history of Spain if i may just uh, go oŸ on a tangent the Basques (70) won’t go through the details of this calculation but i’ll just sketch it very brie¶y, to show you what’s (71) not trying to catch you on anything but i wonder if you’d just comment on it (72) okay predict possible, let’s just put it in there predict possible changes that could, result from
190 Anna Mauranen
(71) is a request for comment in a thesis defence, and although the utterance in itself looks fairly innocent, it is heavily downtoned, including the disclaimer in the beginning. The threat of the question to the candidate is induced by the defence situation, and this is apparently anticipated in the speaker’s hedging. Comparing this to speech in the BNC, the main diŸerence is that the proportion of hedges is clearly less than half (21 /50) in the BNC data, although even here it is the largest individual category. In this sample of general speech, other uses such as particularisers (presents us with just such an opportunity, if we don’t…) and emphasisers (I just can’t aŸord the prescription ), especially with imperatives (Go on just turn it oŸ!), were also relatively large groups, and there were more temporal uses (he just received them like, about half an hour before Mike rang up) as well. The minimisers were epistemic as often as strategic in this sample. The epistemic meanings indicated that the modiªed entity is not di¹cult, important, or dangerous (73, 74), while the strategic use related to oŸers and requests (75, 76). (73) s not, it ’s not his summing up, it ’s just a point of clariªcation to indicate (74) ’t want a lot out of Tesco ’s, it will just save turning out, I don’t mind turning (75) Y A awards. Is that in the way, I ‘ll just move it out of the way. (76) Yes give him a little walk Can you just do something with those budgies? In all, just is used more as a hedge in the academic genres. Since the academic speech is American and the general speech British, the possibility remains that there is a variety factor behind the dissimilarity, but the diŸerent preferences for the functions are clear, despite the fact that in itself, the hedging function overrides the proportion of any other function in both corpora. The epistemic meanings were somewhat more common in the academic samples, which perhaps might be expected, given the context. The strategic uses particularly in the lecture data were mostly limited to discourse re¶exivity. Table 7. Summary of just epistemic
strategic
other
total
42 (21) 44 (22) 22 (11)
34 (17) 36 (18) 20 (10)
24 (12) 20 (10) 58 (29)
100 (50) 100 (50) 100 (50)
MICASE (sample of 100) lectures % (sample of 50) other genres % (sample of 50) BNC sampler spoken % (sample of 50)
“They’re a little bit diŸerent”…
5.3 A little bit The ªnal hedge which was assumed to have an important function as minimiser, particularly as a mitigator, was a little bit, which, with its 188 occurrences in toto was about ten times as frequent as somewhat, although only 15 per cent of the frequency of just (see also Table 7). In comparison with the BNC Sampler (n = 166), it was proportionally twice as common in MICASE. In MICASE, the unit a little bit accounts for virtually all occurrences of bit, and also for a third of little, constituting the biggest three-word cluster associated with little by a wide margin. Its use is predominantly strategic. Most instances are clearly mitigating something negative , either a negative assessment (things get a little bit complicated), an apology of some sort, or as if assuming that there might be resistance to the utterance, like in making a suggestion, or in discourse re¶exive use. Half of the cases are discourse re¶exive, as in examples (77) and (78). We could also include commentary on the context of situation under re¶exivity, as in (79) to (81), where (79) expresses the speaker’s intention, which may or may not succeed, depending on the hearers’ willingness to cooperate. Hedging may make the suggestion more palatable to the hearers while acknowledging the hearer’s role to accept what is proposed. If we include examples of this type under re¶exivity, then the proportion of such examples come to represent more than two thirds of the total. However, these examples would still be included in the strategic category. (77) (78) (79) (80)
and i want to say a little bit about how things are going with the projects, now i wanna spend a little bit of time talking about natural selection… i_ i’m gonna try and make this class go back and forth a little bit. obviously it’s a little bit warmer in this room for me, um than it is for you guys, (81) give you guys some time to to wrestle with it in your mind, and and, understand it maybe at a little bit deeper level, Other strategic examples were not really re¶exive in either sense, but seemed to be mitigating di¹cult or sensitive communication: in (82) the hedge de¶ects a possible interpretation of boastfulness on the part of the speaker while (83) is said in self-defence, minimising the validity of A’s interpretation of B’s act as a potential breaking of rules. (82) now this is a somewhat controversial area. and, which, i’ve been involved with a little bit… (83) A: are you talking to your neighbor about the answers to your quiz? B: she’s not very conªdent about her answers and i’m just give_ helping her out a little bit…
191
192 Anna Mauranen
Typically, these are explicitly dialogic, involving deictic pronouns. A little bit was relatively rare in an epistemic function, minimising the importance (84) of something, or simply indicating (85) that a distinction may not be very clear. The last example could plausibly be given a more strategic reading as well, in which the repeated hedge suggests the personal nature of seeing the eŸects of the colours, diminishing also the threat to hearers who do not see them in the same way. (84) the names vary a little bit but the ideas of what they do are the same. (85) they stand out they come forward toward us a little bit they stand out a little bit from the, from the relatively, cool, bluish purples… In this data, then, a little bit seems to be used strategically to mitigate and to soften. It has a concessive feel: “I know this is not so welcome, but it’s not as bad as it might be”. In the British data the same tendency towards strategic use was evident, but with a distinct bias to negative contexts. Evaluative adjectives modiªed by a little bit were often negative (it ‘s a little bit vague; it goes a little bit odd doesn’t it), or they picked out a negative interpretation, where in other contexts a positive or neutral one would have been possible (86, 87, 88). (86) They’re a little bit hard (87) it probably can look a little bit clinical (88) I would have to be a little bit slimmer, It was also used for mitigating negative evaluations (a little bit heated on occasion; I think it ‘s a little bit late in the day) and in suggestions, in other words, in the basic interpersonal uses of hedges recognised in most work on hedging and politeness (89–91). (89) So I think we ought to see a little bit more of these doctors (90) So perhaps there ‘s a, a little bit of work we can do in er in that sphere (91) Cheese and biscuits? A little bit of that yes, to have afterwards mm?. The implied or explicit negatives in the contexts in the British data were interesting, so I also sampled the ten million word spoken section of the Bank of English, and the ªndings turned out to be very similar; there were a large number of negative heads to a little bit (92–95), and a negative interpretation was picked out of bivalent or potentially positive expressions (96, 97). (92) (93) (94) (95)
what I do did feel was a little bit sad was Cos you can be a little bit awkward at times they do become erm a little bit uncontrollable the things that are perhaps a little bit more di¹cult and complicated to app
“They’re a little bit diŸerent”… 193
(96) become what I would say a little bit ordinary (97) It probably is a little bit early to to to say . Although there were re¶exive uses in the British data as well, the negative ones were more prominent, whereas in the MICASE data the re¶exive use seemed primary. The diŸerence may either be due to the diŸerent genre selection or to the variety, but with the present material that is impossible to resolve. In all, a little bit acted as a single unit, and was the most frequently strategically used item in this selection. Its two major uses were in connection with re¶exive utterances (similarly to just), and in prefacing negative evaluation. At one level it could be said that it mitigates the negative evaluation that follows but on the other hand it seems to be so strongly associated with negativity that it picks out negativity even with potentially bivalent adjectives and makes neutral ones (e.g. colour terms) negative, in other words, it seems to have a negative prosody. Table 8. Summary of a little bit epistemic
strategic
other
total
16 (7) 34 (17) 12 (6)
80 (32) 56 (28) 76 (38)
3 (1) 10 (5) 12 (6)
99 (40) 100 (50) 100 (50)
MICASE lectures % (total of 40) other genres % (sample of 50) BNC sampler spoken % (sample of 50)
To summarise the ªndings concerning all the mitigating hedges discussed in this section, then, despite their commonalities, their individual proªles were diŸerent. Somewhat was rare, and chie¶y used epistemically. It had a negative prosody, thus acting indexically as a forewarning of a negative evaluation. This feature was stronger than in a little bit, which showed similar tendencies, although more clearly in general (British) speech than in the academic (American) data. Just was very frequent, and despite its many meanings was used largely as a minimising hedge, with no clear semantic prosody attached to it. It was divided between epistemic and strategic uses fairly evenly, and in the strategic use an important function was to oŸset the imposition of the speaker’s order on the discourse situation in discourse re¶exive contexts. A little bit shared this function, but was more negative, and overwhelmingly strategic. In frequency it fell between the two extremes. Comparing the frequencies with BNC data, all of the mitigators were proportionally overrepresented in MICASE. The corpus being both small and in the making, the data is not very reliable, but some major trends seemed fairly clear and interesting: while just and a little bit were proportionally somewhat more frequent
194 Anna Mauranen
in MICASE than the BNC (about 1.6 and 2 times, respectively), somewhat occurred three times more often in MICASE. This last ªgure is so large as not to be likely to even out with more data, so that it may well be more characteristic of academic speech than general speech, even though the extra caveat of American vs. British diŸerence still holds.
6.
Conclusion
The hedges discussed in this paper share certain basic functions, most fundamentally adding indeterminacy of one kind or another to an utterance, or, as it has been phrased in this analysis, conveying openness to the utterance, either epistemic, in not ªnalising precise referential meanings, or strategic, in leaving space for interactants’ reinterpretation of or intervention in the discourse situation. Yet the individual hedges had their own proªles on various parameters. In terms of sheer frequency the variation was considerable, with just and sort of/kind of very frequent, somewhat and or so quite rare, and a little bit and or something in between. The distinction between epistemic and strategic uses was maintained with su¹cient ease to warrant its application, even though, not unexpectedly, overlappings and bifunctional cases also appeared. Each hedge appeared in both functions (although or so in the strategic function only in the BNC), but their preferences varied widely: highly epistemic were or so, or something and somewhat, while the most strategic one was a little bit. On the whole, those initially classiªed as vagueness indicators, tended to prefer epistemic uses, and those classiªed as mitigators preferred strategic uses, with somewhat as a clear exception. A genre distinction suggested itself in that the more dialogic genres also tended to have more strategic hedges and the lectures more epistemic hedges. This is interesting in view of the fact that the motivation usually given in support of the overarching face-saving interpretation of hedging in academic contexts hinges upon the idea that epistemic uncertainty must be face-threatening to a writer or speaker. This is not likely to be the case in university lectures, which thus on their part provide evidence in support of an interpretation where epistemic hedging is motivated by needs of an epistemic nature. Another implication of the diŸerent distribution of the hedge functions across genres is that ªnding the same items in diŸerent registers does not imply that they are used in the same meanings. This is important to bear in mind when largescale ªndings on item distribution across genres or registers are interpreted. Other divisions emerged in addition to the intital epistemic/strategic one. Among the unspeciªc category markers, sort of and kind of were employed to introduce new information, and categories presumably not known or clear to the
“They’re a little bit diŸerent”… 195
hearer, often ad hoc descriptions or categories, whereas or something and or so were appended to instances exemplifying categories presumably known to or retrievable by the hearer. Another distinction concerned negativity in the context: a tendency to co-occur with negative expressions, i.e. a negative prosody, was discernible with the epistemic uses of somewhat and less clearly with a little bit. Other hedges showed no negative bias in their epistemic functions. A situational connection with negativity seems to be implied in strategic hedging: hedges appear to be used to oŸset a threat, or a negative reaction or interpretation. Finally, discourse re¶exive expressions tended to co-occur with just and a little bit in particular, and to some extent also with sort of/kind of, but less with other hedges. The MICASE corpus also enables investigation along parameters not touched upon here. For instance, kind of and sort of in MICASE data were studied by Poos & Simpson (1999) from the perspectives of gender and disciplinary diŸerences. Despite earlier claims that hedging is particularly characteristic of women’s speech, they found no clear gender distinction (which may of course be due to the speciªc circumstances of an academic community), but instead a clear division between broad disciplinary divisions (physical and biological sciences vs. social sciences and the humanities) and hedging frequency. In the comparison of MICASE and BNC speech, clear diŸerences emerged in the proportionally much greater frequency of somewhat in MICASE, which could tentatively be attributed to the academic genres, and the much smaller frequency of kind of as a hedge in the BNC, which looked more like a distinction between the British vs. American variety. A third diŸerence seemed to be a more distinct negative prosody that a little bit showed in the British corpora. Clearly, in some cases other interpretations than those given above might be equally plausible if the context of each could be very carefully scrutinised, but the present study remains at a surveying level. To get a more detailed picture of the more ambiguous cases, a close-up study of a few texts (as in Lindemann & Mauranen 2001) or one speaker (as in Poos & Simpson 1999) is a useful approach. As it is, the present ªndings indicate something about the range of functions of each item. The implication of ªnding individual proªles for all hedges is that much more subtle meanings and meaning distinctions can be conveyed by hedges than is suggested by the broad main functions only. Thus, as careful contextual interpretation has lead to discoveries of the diversity of hedging, as for instance in Hyland (1998) or Varttala (2000), a micro-analysis of the kind carried out here, beginning with individual expressions themselves, clearly oŸers an alternative, or complementary perspective on variation and multifunctionality in hedging, which enables the discovery of a speciªc meaning potential associated with each item.
196 Anna Mauranen
References The Bank of English, http://titania.cobuild.collins.co.uk/. Consulted in March, 2000 BNC Sampler corpus 1.0 1999. http://info.ox.ac.uk/bnc Channell, Joanna 1994 Vague Language. Oxford: Oxford University Press. Crompton, Peter 1997 “Hedging in academic writing: Some theoretical aspects”. English for Speciªc Purposes 16: 271–289. 1998 “Identifying hedges: Deªnitions or divination”. English for Speciªc Purposes 17: 303–313. Halliday M. A. K. 1994 An Introduction to Functional Grammar. 2nd ed. London: Edward Arnold. Hübler, Axel 1983 Understatements and Hedges in English. Amsterdam: John Benjamins. Hyland, Ken 1998 Hedging in Scientiªc Research Articles. Amsterdam: John Benjamins. LakoŸ, George 1973 “Hedges: A study of meaning criteria and the logic of fuzzy concepts”. Journal of Philosophical Logic 2: 458–508. LakoŸ, Robin 1975 Language and Woman’s Place. New York: Harper & Row. Lindemann, Stephanie and Mauranen, Anna 2001 “It’s just real messy”; The occurrence and function of just in a corpus of academic speech”. English for Speciªc Purposes 2001: 459–476. Louw, Bill 1993 “Irony in the text or insincerity in the writer? — The diagnostic potential of semantic prosodies”. In Text and Technology. In Honour of John Sinclair, M.Baker, G.Francis, and E.Tognini – Bonelli (eds), 157–176. Amsterdam: John Benjamins. Markkanen, Raija and Schröder, Hartmut 1989 “Hedging as a translation problem in scientiªc texts”. In Special Language: From Humans Thinking to Thinking Machines, C. Lauren and M. Nordman (eds), 171–179. Clevedon: Multilingual Matters. 1997 “Introduction”. In Hedging and Discourse, R. Markkanen and H. Schröder (eds), 1–18. Berlin: Walter de Gruyter. Mauranen, Anna 1997 “Hedging and modality in revisers’ hands”. In Hedging and Discourse, R. Markkanen and H. Schröder (eds), 115–133. Berlin: Walter de Gruyter. 2001 “Re¶exive academic talk: Observations from MICASE”. In Corpus Linguistics in North America, R. Simpson and J. M. Swales (eds), 165–178. University of Michigan Press.
“They’re a little bit diŸerent”… 197
2002
Meyer, Paul 1997
Myers, Greg 1989
“A Good Question”; Expressing evaluation in academic speech”. In Domain-speciªc English; Textual practices across Communities and Classrooms, G. Cortese and P. Riley (eds), 115–140. Frankfurt: Peter Lang. “Hedging strategies in written academic discourse: Strengthening the argument by weakening the claim”. In Hedging and Discourse, R. Markkanen and H. Schröder (eds), 21–41. Berlin: Walter de Gruyter.
“The pragmatics of politeness in scientiªc articles”. Applied Linguistics 10 (1): 1–35. Poos, Deanne and Simpson, Rita C. 1999 A question of gender? Hedging in academic spoken discourse. Paper presented at AAAL, Vancouver. Prince, Ellen, Frader, Joel and Bosk, Charles 1982 “On hedging in physician-physician discourse”. In Linguistics and the Professions. Proceedings of the Second Annual Delaware Symposium on Language Studies, R. di Pietro (ed.), 83–97. Ablex: Norwood, N. J. Salager-Meyer, Francoise 1994 “Hedges and textual communicative function in medical English written discourse”. English for Speciªc Purposes 13:149–170. Simpson, Paul 1990 “Modality in literary-critical discourse”. In The Writing Scholar: Studies in Academic Discourse, W. Nash, (ed.), 63–94. Newbury Park: Sage. 1993 Language, Ideology, and Point of View. London: Routledge. Simpson, Rita C, Sarah L. Briggs, Janine Ovens, and John M. Swales 1999 The Michigan Corpus of Academic Spoken English. Ann Arbor, MI: The Regents of the University of Michigan Sinclair, John M. 1983 “Planes of discourse”. In The Twofold Voice: Essays in Honour of Ramesh Mohan, S. N. A. Rivzi (ed.), 70–89. Salzburg: Universität Salzburg. Stubbs, Michael 1996 Text and Corpus Analysis. Oxford: Blackwell. Varttala, Teppo 1999 “Remarks on the communicative functions of hedging in popular scientiªc and specialist research articles on medicine”. English for Speciªc Purposes 18: 177–200. 2000 “Hedging in a corpus of scientiªc research articles on economics, medicine, and technology; Implications for ESP research and teaching”. Paper presented at ESSE 5; The Fifth Conference of the European Society for the Study of English, Helsinki, 25–29 August 2000.
Interaction in written economics lectures The meta-discursive role of person markers Christina Samson University of Florence
1.
Introduction
For a long time, academic writing has been viewed as impersonal and objective — characterised by lexico-grammatical features such as nominalization and the passive voice — re¶ecting the positivist assumption that academic research is purely empirical and, therefore, more credible if explanation and analysis are allowed to speak for themselves. Eradication of the self, as Hyland (2001) states, has the function of emphasising the sharing of knowledge with the whole community, while stressing that a text can communicate directly to the reader without human mediation. By contrast, recent research on meta-discursive devices in research articles (Meyers 1989; Hyland 1998; Salager-Meyer 1998; Breivega et al. 2002), in student academic writing (Crismore and Farnsworth 1990; Thesen 1997; Ivanic 1998; Tang and John 1999; Hyland 2002), in university textbooks (Hyland 1994, 1999, 2000; Bondi 1999; Samson forthcoming) or in historical economics lectures (Del Lungo 2001, 2002), suggests that academic writing does not only convey an ideational content through the use of impersonal language, but that it gains credibility from writers’ projecting themselves into the text, and displaying commitment to their ideas. However, among the diŸerent academic genres which have been investigated, contemporary written economics lectures, developed into their written form from a spoken version, have received little attention, though they seem to me to be particularly suitable for exploring the projection of the author in the text. Most recent studies have focussed on spoken academic discourse, speciªcally on topic identiªcation (Flowerdew 1994; Hansen 1994), on the role of lexical phrases, asides and anecdotes in the classroom (Strodt-Lopez 1987, 1991; DeCarrico and Nattinger 1988), on schematic phasal patterning (Young 1990, 1994), questions and
200 Christina Samson
answers, repetition and reformulation (Thompson 1997; Bamford 2000). They often, however, seem to neglect the author’s presence in the text. In this paper, therefore, I will investigate how academic economic writers convey their knowledge of economics and construct their written published lectures by adopting a personal stance and projecting themselves in their texts, thus challenging what according to many should be written, detached, decontextualised, and autonomous academic language.1 By means of interpersonal meta-discursive devices — in particular person markers — writers show, on the one hand, authorial presence which plays an important role in securing the correct interpretation of the text and, on the other, academic prestige, in the attempt to construct a successful relationship with their interlocutors, taking into consideration their expectations. As Nystrand (1986) observes, the process of writing consists of elaborating a text in accordance with what the writer assumes the reader knows and expects, and the process of reading is a matter of predicting what the text means according to what the reader assumes the writer’s purpose to be. As a result, written communication depends on what either the writer or the reader assumes the other will do or has done. A central aspect of this relationship is that writers, who are social actors, are aware of the need to ªnd out about each other’s purposes and share the knowledge about how texts signal and represent them. Thus, interaction in the texts may be seen as a result of actions of socially situated writers, who use both the linguistic and the social conventions not only to achieve an increasing social participation in their disciplinary community, but also to persuade the reader into accepting the constructed text (Kress 1989). The examination of person markers, then, serves the purpose of underscoring the functional choices economists/academics make in their written economics lectures relative to their diŸerent needs and to the requirement of adopting diŸerent identities in their texts. On one hand, they may adopt the role of expert/teacher with an expository and explanatory function which characterises written economics lectures as a pedagogic genre. On the other, the writers may use the lecture form to promote their own research results, thus introducing features of an advertisingmarketing genre. This study has the further purpose to show that contemporary written economics lectures are a mixed genre, and that with certain features they subscribe to values distinguishing them from other pedagogic texts within the academic economic community.
2.
Corpus and methodology
The re¶ections presented in this paper are based on some partial results of a qualitative and quantitative analysis of a specialised corpus of ten written economics
Interaction in written economics lectures 201
lectures,2 produced by diŸerent contemporary economists, on various topics of macroeconomics that exemplify academic discourse in this ªeld. All the lectures examined have been expanded, by their authors, from their original spoken version, and have mature students and professionals as their target readers. The lectures have a common macro-structure: that is an introduction, in which the writer announces the direction the lecture will take, a middle, in which the writer develops his theses, by formulating hypotheses and creating model-worlds, and a conclusion. This macrostructure is repeated throughout the lectures in each chapter. All the texts were scanned to produce a specialised electronic corpus of about 130,000 words and analysed by means of Wordsmith Tools. A frequency analysis was conducted to provide quantitative data for the interpretation of the person markers in the written lectures, whereas semantic and discursive functions of the person markers were analysed qualitatively in the texts. An initial analysis indicated that the lectures display mainly three aims: ªrst, to familiarise less professional readers with the norms and assumptions of the discourse community, while updating them on the concepts, the methodology and the practices of the economic discipline; second, to help readers focus their attention on the important points of the issues discussed in the texts in accordance with the norms of pedagogic texts; third, to try, as in research articles written to be read by one’s peers, to emphasize the originality and importance of the writer’s economic research, while promoting oneself and seeking the acceptance and recognition of the scientiªc community constituted by a range of values, assumptions and practices. It may be surprising to ªnd a high display of interpersonal meta-discursive features in an academic genre traditionally considered merely monologic, and in which accepted knowledge is arranged into a coherent form. However, in this paper I seek to establish how, in written economics lectures, person markers have a high frequency and more than one function. They are used to express not only authorial and authoritative prominence, through self-mention, but also to take on diŸerent meta-discursive roles with the function of helping the less expert reader to understand the diŸerent parts of the lectures, in order to reinforce the interactional relationship with the addressee and create a sense of solidarity.
3.
Person markers: a functional choice
Person markers seem to have three main interpersonal meta-discursive functions which overlap with textual ones. They express the writer’s socially deªned persona circumscribed by his/her disciplinary community, they underscore interactional aspects of authorial presence that contribute to the variability in tenor, and they appear to have a promotional academic and professional purpose.
202 Christina Samson
The choice the academic writer makes to be present or absent in the text, by choosing a speciªc person marker, is related to his/her attitude toward the propositional material. It is a conscious choice that contributes to ego involvement, to the degree of intimacy or remoteness, and to the ways the writer wishes to be represented in the text. However, the choice of a speciªc person marker may also be viewed according to Ivanic (1998: 98) as an aspect of the context. Texts simultaneously enact what Halliday (1994) calls the ideational, interpersonal and textual functions of language. In contrast with a conception which denotes it as an autonomous system, language may be viewed as a socio-culturally context-dependent means of communication. If viewed as a dynamic concept, language is always constitutive of social identities, social relations and systems of knowledge, and is socially shaped and shaping (Fairclough 1993). This results in a greater manipulation of the genre conventions with the consequent tendency to mix private intentions with socially recognised communicative purposes (Bathia 1997), a phenomenon widely used in many professional genres. Similarly, texts, which are both pedagogic and present research results, cannot be seen as decontextualised for they re¶ect methodologies, rhetorical strategies and arguments constructed to engage and persuade the readers of the claims that are made. Trying to convince not only less professional readers but also colleague economists, involves deploying disciplinary and genre-speciªc conventions that make the lectures in the corpus into a multi-layered hybrid co-produced by the authors and the readers (Hyland 2000). Therefore, the stance the economists/writers adopt towards their texts indicates a socially deªned persona3 who has acquired authority by speaking and using the codes and the identity of a community member, but it is also related to the writer’s convictions and engagement with the reader. In the corpus, the choice of the ªrst person pronoun corresponds to a multiple discourse identity (Fetzer and Akman 2002) re¶ected in various degrees of authority within the text, where authority means not solely the writer’s possession of knowledge and expertise, or the right to control and command others, but the more speciªc meaning of maker of meaning in Ivanic’s (1998: 12) sense4. A written utterance, in fact, has meaning only if and when it communicates a who and a what. Following Gee (1999), who means a kind of person one is seeking to be and enact whereas what is a socially situated activity that an utterance helps to constitute, as can be seen in examples (1) and (2): (1) I will follow a modeling strategy that implies dropping the representative ªrm as used by Marshall. Thus I am asking about the importance of what Marshall called the “eddies” when one is thinking about events that will occur some time in the future. First I will describe a simple example of a model that tries to come to grips with some of the issues of modeling time
Interaction in written economics lectures 203
explicitly. Then I will turn to empirical work to underscore the importance of simultaneous consideration of these diŸerent factors. (Diamond) (2) That we present alternative theories as honestly as we can does not imply that we are theoretical wimps. We believe that most (not all) current theories do capture important aspects of reality; we do not believe in monocausal or monodistortion accounts of ¶uctuations. We believe that eclecticism in the pursuit of truth is no crime; we are sure, however, that our preferences, which are obviously re¶ected in our research, will be clear to the careful reader. (Blanchard and Fischer) The examples show how the writers textualise their work as a contribution to the ªeld while constructing themselves as competent members of the discipline; the examples also show by the use of the person marker how the academic authors are constrained by complex interpersonal relationships, diŸerences in status and power, which entail their identity being constructed by their membership of, and identiªcation with, the values and practices of one or more communities (Ivanic 1998: 83). The economic discourse community can thus be considered a socio-rhetorical network, which is formed in order to work toward a set of common goals (Berkenkotter and Huckin 1995). In this community the choice of personal reference may be viewed as a clear indication of the academics’ aim to balance objective data, hypothetical worlds, and a sense of disciplinary solidarity while in¶uencing the perspective from which their texts should be interpreted. Person markers (italicised) are ªrst person personal pronouns characteristically used in a situational context, in that they ‘shift’ according to the situation, and point, deictically speaking, to ‘the one who is speaking’ and also ‘the one who is reading’ — I and we (Wales 1996). For the purpose of my investigation of the interactional use of person markers, I follow overall Hyland’s deªnition, which deªnes person markers as items of a subcategory of interpersonal meta-discourse which contribute to the writer-reader relationship (Hyland 1994, 2000). As shown in Table 1, the lectures in the corpus present a high frequency of the ªrst person pronouns — I and we — revealing the decision on the part of the writers to take a stance and adopt diŸerent roles in their texts. A plural ªrst person pronoun can have either inclusive or exclusive semantic reference. An inclusive we includes both writer and reader, whereas the latter excludes the reader. More speciªcally, I and exclusive we (4,408 I vs 720 we) may often overlap in indicating: (1) a clear choice to establish an authoritative self as a member of the economic community; (2) the writer’s prominent position towards less expert readers; (3) the roles adopted in the dialogue which help enforce asymmetry; (4) promotion of research and the writer’s self-image; (5) that written economics lectures form a mixed academic genre.
204 Christina Samson
Table 1. Frequency of person markers type of person marker
frequency
I me my we our Total
4,408 95 84 720 90 5,397
In Section 4, I will start by analysing the roles the writers/economists adopt in the construction of their lectures; then I will investigate how the same person markers may have diŸerent or overlapping functions in accordance with the writer’s purposes, thus often becoming the expression of personal intentions. Consequently, the use of speciªc person markers may be viewed as a demonstration of research prominence whilst signalling a complex relationship between particular instances of language use in dynamic academic/professional communication.
4.
Person markers: instances of authorial roles
The presence or absence of explicit author reference is a conscious choice made by writers who want to adopt a particular stance and have a particular role in their text. In the corpus, person markers vary according to the sections of the lecture, to the function and role5 the writer wants to adopt whilst interacting with the reader, and according to the degree the reader is to be included in the text. The focussing on the speciªc needs and requirements of the discipline emphasizes that academic writing should not be considered monolithic or shaped in a mechanical way; on the contrary, as documented in several instances of genre-mixing, there is an extensive restructuring of boundaries between discursive practices (Bathia 1997). This feature is also found in these economics lectures which are characterised by a wider use of interpersonal meta-discursive devices to promote the academic’s research as underlined by the roles adopted in the corpus which I have broadly classiªed as: ‘author’, ‘organiser’ and ‘guidance’ role.
4.1 Author role Of the roles adopted by the writers, the author role is amongst the primary ones in the corpus. The writer adopts this role in order to provide the reader with the knowledge background which demonstrates the writer’s research experience in the ªeld, before presenting his/her own research.
Interaction in written economics lectures 205
As in the introductions of research articles, in the lectures, the writers ªrst deªne the ªeld of study and then establish a niche for their research in the integral part of the text. In doing so, they claim their authority as a source of knowledge, positioning themselves asymmetrically in relation to their readership. However, to mitigate their superiority and show deference towards the scientiªc community members, the authors maintain a colloquial style, often by expressing a personal view on established facts. The use of a colloquial style is underscored by the presence of I, a signal of a restricted referential ªeld since it refers to the self and, more speciªcally, has come to be a prototypical index of subjectivity (Wales 1996); I signals also the responsibility for a critical claim which is simultaneously a form of hedging, of what may be a face-threatening claim, as in (3): (3) I have been researching the inadequacies of conventional approaches of the modelling of time since 1968. My dissatisfaction with treatments of stability led me to think about price adjustments in real time, with individuals aware that they are partaking in a process in real time. […] (Diamond) This example illustrates how reference to the writer’s previous research is emphasised by the possessive determiner my, with reference to I, whilst declaring his dissatisfaction with the inadequacies of conventional approaches of the modelling of time since 1968. Me is used when the writer wants to reconsider the question because he holds his own opinion on the topic. By using the ªrst person pronoun with the cognitive verb think, the writer on one hand weakens his claim, yet allows the reader to judge and criticise his statement. In (4), the writer introduces his topic as if writing a research article, emphasising the centrality and reason for his current research by detaching himself from the topic and beginning with the impersonal there has been a ¶urry of activity in order to state established facts: (4) […] over the last ªfteen years, […] there has been a ¶urry of activity […] particularly the life-cycle theory of Modigliani and Brumberg (1954, 1979) which is the basis for essentially all modern research on consumption and saving. In these chapters I try to tell the story of the most recent burst of research. […] (Deaton) Impersonalisation is usually seen as a contribution to the objectivity of scientiªc discourse; however, it may be a face-saver to protect the writer from his peers. Such an attitude is underlined, in example (4), by a ‘non- integral’ citation (particularly the life-cycle theory of Modigliani and Brumberg) 6which shows, on one hand, the
206 Christina Samson
writer’s familiarity with the research ªeld, and, on the other, helps to create a personal research space for the writer. In the hedged statement I try to tell the story he addresses his peer readership more directly. In (5) the academic writer once again deploys his knowledge of the ªeld, referring to an integral citation of Marshall, acknowledging his work, showing deference to the whole scientiªc community. While sharing Marshall’s conception of time, he acknowledges that this may nevertheless result in problems. By using I the writer takes up a personal stance, thus hedging a claim which may be seen as the expected one (5) In the Preface to the ªrst edition of his Principles of Economics, Alfred Marshall refers to the “element of Time” as “the centre of the chief di¹culty of almost every economic problem” (1948, p.ii). I share Marshall’s view of time as a source of di¹culty (Diamond).
4.2 Organiser role In exploring the lectures of the corpus, I noticed that the ªrst person pronouns, singular or plural, mainly occur when academic writers, in the introduction of a chapter or of a section of their lectures, assume the role of ‘organiser’. In such a role writers have to decide how to structure the material by outlining and organising it, in order to achieve their communicative purposes. In the role of authorial prominence, found mainly in the introduction and middle sections of the lecture, I collocates with verbs indicating both the organisation of the issue presented, thus creating a text frame, and a commitment to what the reader expects to read ahead in the text. In these sections, the person markers serve also as a textual meta-discursive device, which often overlaps with the interpersonal function to explain hypotheses, models and illustrations. For this purpose, the person markers co-occur with verbs referring to the analysis of non-linear texts such as equations, charts, ªgures, etc or with verbs that direct the attention of the reader to what follows in the text. The most frequent verbs found in these two sections are listed in Table 2: Table 2. Verbs in introduction /middle sections analyse assume consider demonstrate develop discuss
examine explore extend focus include introduce
present return see show start study
Interaction in written economics lectures 207
As I found in a previous study (Samson forthcoming), the verbs occur mainly in the present tense. There were no instances of present progressive, which was to be expected since the progressive forms would introduce a narrative element in the academic presentation that would move the reader’s attention away from what is argued or claimed about the performance of the activity itself. The awareness on the part of the academic/writer of how to employ diŸerent personal pronouns to increase or decrease the reader’s inclusion in the text is exempliªed by the use of the person marker we in the corpus. The frequent use of exclusive we reinforces the writer’s power when it refers to the author in the lectures, as can be seen in Blanchard and Fischer’s example (6). (6) In this chapter and the next we focus on the fundamentals of consumption and capital accumulation in dynamic nonmonetary equilibrium models. We introduce basic models — in this chapter, the Ramsey inªnite horizon optimizing model, and in the next, overlapping generations models with ªnite horizon maximizers — and begin to analyse economic issues such as how much interest rates aŸect savings and whether the choice between tax and deªcit ªnancing aŸects capital accumulation (Blanchard and Fischer) The authors use we to show that they take full responsibility for their utterances. They tell the reader what they will focus on, and what their goals will be. We like I, in the previous examples, is related to the function of describing the writer’s procedural choice at the beginning of a chapter or section or at the end when announcing what follows. The organiser role of the academic/writer is reinforced, as in example (7), by the co-occurrence and the repetition of the future tense will indicating the intention of the writer to carry out certain activities. (7) In these lectures I will examine how time is modeled in various economic analyses. My focus will be on the modeling of equilibrium, particularly equilibrium with many economic agents. I will present a leisurely tour through some economic analyses, with an eye on their treatment of time. The ªrst lecture considers models of a single industry; the second, models of an entire economy. (Diamond) Will co-occurs with the verbs examine, focus, present, which the reader expects to be found in the text. Thus there is a close relation between the agent and the main predicate, which predicts a future completion of the activity. The writer reinforces his commitment to the reader and to the text by specifying the content of the two lectures he will focus on — models of a single industry and models of an entire economy — and by using the two textual meta-discursive sequencers (Tadros 1994) — ªrst and second — which show the reader how the parts of the text are related to
208 Christina Samson
one another, and help to construct an appropriate mental representation of what is said in the reader’s memory. The present tense refers to an action taking place in the immediate context of interaction. Also the grammatical and lexical repetition — as in I will examine, I will present, models — has two functions in the lecture: it enables the reader to process the propositional content more easily, and it signals the economist’s familiarity with the linguistic conventions of the discourse community for giving a reasoned account of one’s thinking.
4.3 Guidance role Another role that may be viewed as highly interpersonal, although the writer maintains his asymmetrical, expert position, is the guidance role. This role is characterised by the usage of inclusive we, with 1st person verbs, in the middle of the lecture when the topic is summarised. It demonstrates personal conªdence based on the author’s command of the arguments as when the writer develops his/her theses by formulating hypotheses and by creating model-worlds. In example (8) the person marker we is clearly inclusive as the writer previously refers to you, the reader. (8) I started by reminding you of the familiar atemporal short and long-run Marshiallian models. I then introduced a model set in real time. As we will see in a moment, this is true of US industries. (Diamond) The author ªrst draws the reader’s attention to the points he has discussed — the short and long-run Marshallian models followed by a real time model — and arrives at the decision of what he will examine next. The writer’s prominent role is rea¹rmed by the use of the 1st person singular I followed by a switch to we in order to create a sense of togetherness with the reader. We is followed by the verb see, which signals the writer’s role to help the reader to see something relevant in the text. By using inclusive we, and involving both writer and reader in the same activity, the writer presupposes that the reader has a certain background knowledge and ability that will allow him to follow the arguments. The use of inclusive we can therefore be seen as a tool to shorten the distance between writer and reader and stress solidarity between the two as when, for instance, the author develops and explains the concepts he will use. However, there are instances, as in example (9), in which the use of we is not clear-cut: (9) […] I include EBRD forecasts of output for 1996 when available, so that Fig. 1.1 gives the behaviour of output […] GDP is normalized to be equal to 1 in year 0, the year before transition. With these preliminaries out of the way, we return to the behaviour of GDP as shown in Fig. 1.1 […] (Blanchard)
Interaction in written economics lectures 209
Especially when a single author has used I while explaining a particular model, the sudden switch to the plural may indicate commonality with the reader, but it may simultaneously, as Pennycook (1994) observes, be interpreted as a claim of authority. Example (9) suggests how the writer can reduce his personal intrusion in the text while emphasizing the importance that should be given to the procedure itself. By referring to himself, as I, the writer guides the reader to an explanation of his GDP thesis. Signiªcantly, he uses present tense verbs (include, return) which indicate immediate action and presuppose visual perception when the reference is to a ªgure or a non-linear text. In example (9) the ªgure is referred to by a noun phrase (‘Fig. 1.1’), collocating with a textual meta-discursive marker — as shown in Fig. 1.1- (cf. Samson forthcoming). In other instances the academic writer seems to seek agreement and cooperation from his readers by using let us before starting to explain a model world as in extract (10). (10) Let us consider this industry under the assumption that the demand curve has a multiplicative factor that follows a sine wave (Diamond). Let us behaves like a pragmatic particle with the illocutionary force of a polite request. In the corpus it functions as a device to involve the reader and it strategically fronts a clause where the writer explains what he will discuss next.
5.
Person markers and self-mention: a promotional role
Clearly, from the analysis of the corpus, person markers appear to be a means by which academics writers display various identities in diŸerent parts of the lectures. Person markers are a powerful signal of self-mention, a rhetorical strategy to demonstrate personal contribution to the economic community and to establish a claim to achieve a recognition of academic priority. In the corpus, the economists frequently cite deªnitions by previous researchers — not only in the preface or in the introductions, but also in the middle sections — and adopt personal stances with the clear intent of demarcating their work from that of others in order to emphasise their innovative contribution to the discipline. In (11), (12) and (13) the authors refer to arguments and methodologies that are established norms of the discipline in order to signal their subjective evaluation (I contrasted; I think this description is basically right) and in order to relate their work to that of their colleagues: (11) Eric Lundberg (1937) devotes considerable attention to the di¹culty of deªning a period for sequence analyses. He makes it clear that his “day” is
210 Christina Samson
not a chronological day; so while there is discussion of alternative lags, there is not an empirical statement about applicability. In my view, continuous-time modelling of this process is conceptually cleaner, but has considerable mathematical di¹culties, especially when diŸerent factors have diŸerent lag structures and diŸerent speeds of response (Diamond) (12) It is unlikely that anyone would have noticed, but in the ªrst lecture I avoided the word “dynamic”. Instead, I contrasted explicit-time and atemporal models, implicitly making a case for both of them, one for theoretical analysis, the other for exposition and applications. This is in contrast with the way that the term “dynamic” is often used. (Diamond) (13) I see the transition as being shaped by two main mechanisms. The ªrst is reallocation. A typical description of what happened at the beginning of transition is that price liberalisation and the removal of subsidies triggered a collapse of state ªrms and that growth in the new private sector was simply insu¹cient to take up the slack. I think that this description is basically right (Blanchard). The persuasive use of self-mention overlaps with the organiser role when the writers use I or we to insist on their contribution to the ªeld; cf. I discussed, I assumed, we identiªed, we noted, I focussed, I argued, I looked in examples (14), (15) and (16). (14) I discussed the optimal contract written by an entrepreneur who raises capital from an investor (or set of investors). I assumed that the entrepreneur obtained signiªcant (private) beneªts from a running a ªrm […] (Hart) (15) We identiªed lots of reasons why ªrms are slow to change prices. And we noted that there are generally many diŸerent prices at which homogenous goods are available (Diamond) (16) I focussed on the U-shaped adjustment of output and unemployment during transition. I argued that the two main forces shaping transition were reallocation […]. I looked in particular at the interaction between restructuring and labour market conditions (Blanchard). The person markers may be seen also from the utilitarian perspective, that how the text is constructed is useful in promoting one’s academic position. This strategy can be stressed by the use of possessive adjectives in combination with nouns such as point, attempt, models as in examples (17), (18), (19), which serve to highlight the writer’s contribution to the academic discipline:
Interaction in written economics lectures
(17) My point is that the process of selection of the “right” atemporal model involves thinking about the resource allocation process over time, and that some of that thinking may be done better in an explicit model (Diamond) (18) Our models rarely generate such simple messages. The analysis of transition is very much an analysis of the best-second, and comes with the typical attendant ambiguities (Blanchard) (19) These lectures re¶ect my attempt to think about behaviour of an economy over time… They also re¶ect my attempt to think coherently about micro and macro… (Hart) Self-mention has not only the function to demonstrate that one belongs to the academic community and the ‘cutting edge’ of one’s research. Another function is rhetorical timing (Berkenkotter and Huckin 1995); the speaker chooses a topic at a moment when it can be expected to be well received by the academic community because of its newness and the background knowledge it presupposes. This feature distinguishes written economics lectures from their oral form and from economics textbooks, and indicates that they should be classiªed as a mixed academic genre. In extract (20), for instance, it is understood that the economist thinks this is the appropriate moment to discuss his research since his conclusions could be irrelevant in few years. (20) In these chapters I try to tell the story of the most recent burst of research. …I do not know whether now is the best time to try. …But the story as it now stands is a good one; a great deal has been learnt that we did not know before, …So the tale is worth telling, even at the risk that the conclusions will be transparently wrong only a year or two from now. (Deaton) The example shows that these lectures have both a pedagogic purpose and the purpose of promoting the writer’s own research. The genre can therefore be described as a mixed one. Similarly in (21) the writer explains the reasons and background knowledge for his research. The future changes of an economic transition in Eastern Europe represent a challenge which he is going to take up: (21) Transition in Central and Eastern Europe has led to a U-shaped response of output, that is, a sharp decline in output followed by recovery. Six years after the beginning of transition, most of the countries of Central Europe now seem ªrmly on the upside. Most of the countries of Eastern Europe are still close to the bottom of the U […]. One of the challenges facing those working on transition, however, is whether they can convincingly explain the diŸerences between Central Europe, Eastern Europe, and China. I shall take up this challenge as I go along (Blanchard)
211
212 Christina Samson
Furthermore, in (22), the two economists Blanchard and Fischer promote their own research strategy by stating that alternative strategies have been unsuccessful: (22) The Keynesian framework embodied in the “neoclassical synthesis”, which dominated the ªeld until the mid-1970s, is in theoretical crisis, searching for microfoundations; no new theory has emerged to dominate the ªeld, and the time is one of explorations in several directions with the unity of the ªeld apparent mainly in the set of questions being studied. […] We believe that looking at their eŸects as rising from deviations from a wellunderstood benchmark is the best research strategy. Alternative strategies that have started squarely from a diŸerent benchmark have for the most part proved unsuccessful (Blanchard and Fischer)
6.
Concluding remarks
From this investigation, it seems clear that written economics lectures, as any other academic discourse, do not occur in a social vacuum. On the contrary, they always have to be understood in the context of their discipline. The use of person markers is a signiªcant meta-discursive device. Self-mention can for instance be used to build authorial authority, promote one’s research and self, and persuade readers, and thus it fulªls needs within the disciplinary community. The distribution of speciªc person markers in diŸerent parts of the lecture indicates that any decision on the part of the writers is linked, on one hand, to the necessity to conform to the norms of the discipline in order to be accepted as a member of the academic community; on the other, the person markers are a strategy writers use to appear more prominent in the discourse or to adopt a stance towards their texts and readers. If addressing peer readers, the function of selfmention will be mainly that of self promotion, of persuasion and of boosting one’s credibility and authoritativeness. The use of I (or my) to refer to one’s research procedures act as a means of promoting one’s research. When addressing less knowledgeable readers, the writer uses I and inclusive we to persuade the addressee and to render the text more understandable and interesting through the use of a more colloquial style. The adoption of diŸerent personal pronouns reveals that academics are well aware of how to use rhetorical strategies to negotiate their knowledge while maintaining an asymmetrical position. In their diŸerent roles, writers/academics not only help readers through the text by taking them on a sort of ‘tour’, but indirectly promote themselves. The choice of a speciªc personal pronoun may be seen as a deliberate decision, on the part of the writers, to promote their research and results, a feature which helps to
Interaction in written economics lectures 213
diŸerentiate written economics lectures from the other academic genres. The conscious choices writers/economists/academics make in their text therefore do not merely re¶ect the age and position of the writers, as claimed by Luukka and Markkanen (1997), but are more likely a result of the mixture of private intentions, institutional requirements of generic academic conventions. Written economics lectures may be viewed, then, not as empirical, impersonal and objective texts or as purely pedagogic tools, but they may be classiªed as a mixed genre, in which the writers show a certain freedom to be innovative by exploiting resources belonging to the research article genre and the economics textbooks genre within the broad framework of the specialist economics community.
Notes 1. By contrast, spoken language has been characterised as ‘involved’ and context-dependent, that is as aŸecting and aŸected by the social relations of the interlocutors (Chafe 1982; Tannen 1985). 2. The written economics lectures analysed are included in: Blanchard, O. (1997). The Economics of Post-Communist Transition, Oxford: Oxford University Press. Blanchard. O. and Fischer S. (1997). Lectures on Macroeconomics. London: MIT Press. Deaton, A. (1992). Understanding Consumption. Oxford: Clarendon Press. Diamond, P. A. (1994) On Time. Lectures on Models of Equilibrium. Cambridge: Cambridge University Press. Hart, O. (1996). Firms, Contracts, and Financial Structure. Oxford: Clarendon Press.1. 3. Cherry (1988) distinguishes persona from ethos in writing, with ethos referring to the personal characteristics e.g. funny, interesting, intelligent, etc., that a reader attributes to a writer based on textual evidence; whereas persona refers to both the personality and roles which a writer adopts while producing a piece of writing. 4. Ivanic (1998) takes up Cherry’s (1988) work in exploring the phenomenon of selfrepresentation in writing and focuses on the societal and discourse roles of the textual selves of her participants. Tang and John (1999) add a third role which they call genre role. Genre roles are speciªc to a particular genre within the discourse community. 5. In this paper I refer to Tang and John’s (1999) taxonomy of the various degrees of diŸerent genre roles assumed by a writer in his/her text. However, because the data of this corpus diŸers from theirs, reference to their classiªcation is generic and has been adapted to suit the written lectures under exploration. 6. Citations can be integral and non-integral (Swales 1990). An integral citation contains the name of the cited author as part of the text, whereas a non-integral citation shows the cited author in parentheses or in a footnote or endnote.
214 Christina Samson
References Bamford, Julia 2000 “Interactivity in academic lectures: the role of questions and answers”. In Dialogue Analysis VII: Working with Dialogue, M. Coulthard, J.Cotterill and F. Rock (eds), 138–159. Tübingen: Max Niemeyer. Bathia, Vijay. K. 1997 “Genre mixing in academic introductions”. English for Speciªc Purposes, 16 (3): 181–195. Berkenkotter, Carol and Huckin, Thomas 1995 Genre Knowledge in Disciplinary Communication: Cognition/Culture/Power. New Jersey: Erlbaum Associates. Bondi, Marina 1999 English Across Genres: Language Variation in the Discourse of Economics. Modena: Fiorino. Breivega Kjersti R., Dahl, Trine and Fløttum Kjersti 2002 “Traces of self and others in research articles. A comparative pilot study of English, French and Norwegian research articles in medicine, economics and linguistics”. International Journal of Applied Linguistics, 12 (2): 218–239. Chafe, Wallace 1982 “Integration and involvement in speaking, writing, and oral literature”. In Spoken and Written Language, D. Tannen (ed.), Norwood, N.J: Ablex. Cherry, Roger 1988 “Ethos vs persona: self-representation in written discourse”. Written Communication, 5: 251–276. Crismore Avon, and Farnsworth, Rodney 1990 “Metadiscourse in popular and professional science discourse”. In The Writing Scholar. Studies in Academic Discourse, W. Nash (ed.), 118–136. London: Sage. DeCarrico Jeannette and Nattinger, James 1988 “Lexical phrases for the comprehension of academic lectures”. English for Speciªc Purposes, 7: 91–102. Del Lungo, Gabriella 2001 “Interactional aspects in Marshall’s lectures to women”. Quaderni del Dipartimento di Linguistica, Università degli Studi di Firenze, 11: 157–170. 2002 “The negotiation of academic knowledge in nineteenth-century lectures on economics”. In Con¶ict and Negotiation in Specialized Texts, Selected Papers of the 2nd Cerlis Conference, M. Gotti, D. Heller and M. Dossena (eds), 319– 334. Bern: Peter Lang. Fairclough, Norman 1993 “Critical discourse analysis and the marketization of public discourse: the universities”. Discourse & Society: 130–165. Fetzer, Anita and Akman, Varol 2002 “Contexts of social action: guest editors’ introduction”. Language & Communication 22: 391–402.
Interaction in written economics lectures
Flowerdew, John 1994 Academic Listening. Research Perspectives. Cambridge: Cambridge University Press. Gee, James Paul 1990 Social Linguistics and Literacies: Ideology in Discourses. London: Farmer Press. Halliday, M. A. K. 1994 An Introduction to Functional Grammar. 2nd ed. London: Edward Arnold. Hansen, Christa 1994 “Topic identiªcation in lecture discourse”. In Academic Listening. Research Perspectives, J. Flowerdew (ed.), 131–145. Cambridge: Cambridge University Press. Hyland, Ken 1994 “Hedging in academic writing and EAP textbooks”. English for Speciªc Purposes, 13 (3): 239–256. 1998 “Persuasion and context: the pragmatics of academic discourse”. English for Speciªc Purposes 30 (4): 437–455. 1999 “Talking to students: Metadiscourse in introductory coursebooks”. English for Speciªc Purposes 18 (1): 3–26. 2000 Disciplinary Discourses. Social Interaction in Academic Writing. Harlow: Longman. 2001 “Humble servants of the discipline? Self-mention in research articles”. English for Speciªc Purposes 20: 207–226. 2002 “Authority and invisibility: authorial identity in academic writing”. Journal of Pragmatics 34: 1091–1112. Ivanic, Roz 1998 Writing and Identity: The Discoursal Construction of Identity in Academic Writing. Amsterdam: Benjamins. Kress, Gunther 1989 Linguistic Processes in Sociocultural Practice. Oxford: Oxford University Press. Luukka, Minna-Riitta and Markkanen, Raija 1997 “Impersonalisation as a form of hedging”. In Hedging and Discourse, R. Markkanen and H. Schroder (eds), 168–207. Berlin: W. de Gruyter. Meyers, Greg 1989 “The pragmatics of politeness in scientiªc texts”. Applied linguistics 4: 1–35. Nystrand, Martin 1986 The Structure of Written Communication. Orlando: Academic Press. Pennycook, Alastair 1994 “The politics of pronouns”. ELT Journal 48 (2): 173–178. Salager-Meyer, Françoise 1998 “Hedges and textual communicative function in medical English written discourse”. English for Speciªc Purposes 13: 149–170. Samson, Christina forthcoming “Advance labelling in macroeconomics textbooks”. In Dialogue within Discourse Communities: Metadiscursive Perspectives on Academic Genres, M. Bondi and J. Bamford (eds), Tubingen: Max Niemeyer.
215
216 Christina Samson
2002
“Negotiating academic knowledge: The use of person markers in contemporary written economics lectures”. Quaderni del Dipartimento di Linguistica. Università di Firenze 12: 183–195. Strodtz-Lopez, Barbara 1987 “Personal anecdotes in university classes”. Anthropological Linguistics 29 (2): 194–258. 1991 “Tying it all in: asides in university lectures”. Applied Linguistics 26 (6): 117– 140. Swales, John 1990 Genre Analysis. Cambridge: Cambridge University Press. Tadros, Angela 1994 “Predictive categories in expository texts”. In Advances in Written Text Analysis, M. Coulthard (ed.), 69–81. London: Routledge. Tang, Ramona and John, Suganthi 1999 “The ‘I’ in identity: Exploring writer identity in student academic writing through the ªrst person pronoun”. English for Speciªc Purposes 18: S23-S39. Tannen, Deborah 1985 “Relative focus on involvement in spoken and written discourse”. In Literacy, Language and Learning, Olson D., Torrance N. and A. Hildyard (eds), 172–194. Cambridge: Cambridge University Press. Thesen, Lucia 1997 “Voices, discourse, and transition: In search of new categories in EAP”. Tesol Quarterly 31 (3): 487–511. Thompson, Susan 1997 “Why ask questions in monologue? Language choice at work in scientiªc and linguistic talk”. In Language at Work. Selected papers from the Annual Meeting of the British Association for Applied Linguistics. University of Birmingham, Hunston S. (ed.), 137–150. Clevedon: Multilingual Matters. Wales, Katie 1996 Personal Pronouns in Present-Day English. Cambridge: Cambridge University Press. Young, Lynne 1990 Language as Behaviour, Language as Code. Amsterdam: Benjamins. 1994 “University lectures – macro structure and micro features”.In Academic Listening. Research Perspectives, J. Flowerdew (ed.), 159–176. Cambridge: Cambridge University Press.
Part III
Text and information structure
Using non-extraposition in spoken and written texts A functional perspective Gunther Kaltenböck University of Vienna
1.
Introduction
Structurally, non-extrapositions are related to it-extrapositions, their syntactic counterpart, with which they form a syntactic paradigm or “thematic system” (Huddleston 1984), as illustrated in (1). (1)
non-extraposition That John went to Paris is surprising What John will do in Paris is obvious Living in Paris is fun To ªnd a job will be imperative For John to travel to Paris is imperative
it-extraposition It is surprising that John went to Paris It is obvious what John will do in Paris It is fun living in Paris It will be imperative to ªnd a job It is imperative for John to travel to Paris
The structural relationship between the two constructions is such that non-extraposition, which preserves the canonical SVC order, is usually taken as the more basic form, from which the other is derived by shifting (extraposing) the subject clause to the right and inserting the ‘dummy pronoun’ it (cf. e.g. Huddleston 1984: 451, Quirk et al. 1985: 1391–1393).1 While the syntax of itextraposition and non-extraposition, notably their respective derivations, have received considerable attention especially in transformational grammar (cf. e.g. Rosenbaum 1967; Huddleston 1971; Emonds 1972, 1976), their communicative function in texts has largely been disregarded.2 It is clear, however, that, despite their structural relationship and logico-semantic synonymy, the two constructional types behave quite diŸerently when used in texts so that they are not usually exchangeable in a given context. This functional diŸerence is already indicated by
220 Gunther Kaltenböck
frequency evidence, which shows that non-extraposition occurs far less frequently than its extraposed counterpart and therefore has to be taken as the statistically marked construction of the two (cf. e.g. Givón’s 1995 criteria of markedness). The aim of this paper is to explore the use and functional properties of precisely this marked variant, non-extraposition, and compare it, where relevant, with the ªndings for unmarked it-extraposition. The discussion is based on a corpus study of 217 instances of non-extraposition and 1,701 examples of it-extraposition which were retrieved from the ICE-GB corpus, the British component of the International Corpus of English. This one million word computer corpus comprises a total of 500 texts with an equal share of spoken and written language.3 The paper is divided into two main parts. After a brief delimitation of the class of non-extraposition in Section 2, I will ªrst of all discuss the occurrence and distribution of non-extraposition in the corpus (Section 3). The second part (Section 4), provides a functional perspective of non-extraposition with a close analysis of the relative length of subject and matrix clauses as well as the information status of the clausal subject. This will allow us to establish the main communicative properties of this constructional type, which in turn oŸers a functional explanation for the quantitative ªndings in Section 3.
2.
Deªning the class
As can be seen from the examples in (1), the class of non-extraposition includes various syntactic subtypes depending on the form of the non-extraposed clausal subject, which may be a that-clause, wh-clause, -ing clause, to-inªnitive, or, rarely, a for/to-inªnitive. Some restrictions, however, apply to -ing clauses because of their inclusion of both nominal and verbal forms (cf. e.g. Huddleston 1984: 312–317, Quirk et al. 1985: 1290–1292 for a discussion of the cline linking the two forms). It is necessary to exclude clear-cut nominal instances since their ‘extraposed’ counterpart is not, in fact, an it-extraposition but a right-dislocation (cf. It is a nuisance, the decorating of the room). The procedure adopted for the present study was a restrictive one, which excludes not only those fairly obvious cases of nominal gerunds where the -ing form is preceded by an adjective, article or other forms of determiner, or postmodiªed by an of-genitive, but also cases of single (i.e. unexpanded) -ing forms; e.g. Overtaking can be a very dangerous manoeuvre (S2A-054-205). Similarly, it could be argued with what-clauses that, strictly speaking, only the interrogative type (e.g. What to put in its [=the poll tax] place has been debated long and hard (S2B-030–120)) is ªrmly part of non-extraposition, whereas nominal relative clauses (e.g. What you’re doing is sensible and right (S2B-003–43)), because of their closeness to noun phrases, are borderline cases which in their ‘extraposed’
Using non-extraposition in spoken and written texts 221
version could be classiªed as right-dislocations rather than it-extrapositions proper. However, given the formal criterion of a clausal status, all types of what-clauses were included in the class of non-extraposition. When-clauses, too, have a special status because of their intermediary position between subject and adverbial clauses, which makes it di¹cult for them to take subject position (cf. Quirk et al. 1985: 1392). Not surprisingly, therefore, no such cases have occurred in the corpus. With regard to the matrix predicate, non-extraposition is restricted in that the predicate may not contain another subordinate clause as a complement (cf. (2a)) since this makes extraposition impossible (cf. (2b)) (cf. also Erdmann 1987: 40, Collins 1994: 14). (2) a.
That he continued working hard after all the disappointments shows that he was extremely motivated b. *It shows that he was extremely motivated that he continued working hard after all the disappointments
With non-extraposition being essentially deªned by its structural relationship to a corresponding extraposed version this also excludes identifying (equative) constructions such as in (3), which express a semantic relationship of identiªcation between subject clause and complement. (3) a. To say actors are childlike is to pay them a compliment (W2B-004-2) b. What I didn’t like uh was leaving my mum (S1A-076-98) c. How you fashion them of course is the big big problem (S1A-089-23) Unlike attributive (predicational) non-extraposition, which speciªes a relationship between an entity (subject clause) and some attribute that is ascribed to it (matrix clause), the above constructions do not allow extraposition of the subject clause but can be ‘reversed’ instead (cf. Halliday 1994: 119–124): To pay actors a compliment is to say they are childlike, Leaving my mum was what I didn’t like, The big big problem is how you fashion them. Examples (3b) and (3c) are thus easily identiªable as instances of WH-clefts (basic pseudoclefts). With an NP in matrix predicate position there is, however, the potential of ambiguity between an identifying and an attributive reading. This is illustrated by example (4a), which allows reversing of subject and complement (cf. (4b), identifying construction) as well as extraposition (cf. (4c), attributive construction). (4) a.
Just how those and later funds were spent is a key point in the investigation (W2C-010–65) b. A key point in the investigation is how those and later funds were spent c. It is a key point in the investigation how those and later funds were spent
222 Gunther Kaltenböck
In such cases only those constructions were included in the class of non-extraposition where an extraposed counterpart was structurally possible, i.e. independent of contextual considerations. While the availability of the extraposed version is indeed the deªning criterion for the extraposition/non-extraposition paradigm, it will be shown (cf. Section 4) that the question of structural choice within this paradigm applies only on a more abstract level: in actual use this potential choice is heavily constrained by functional factors.
Occurrence in the corpus
3.
Before turning to a detailed functional analysis in Section 4, let us look at frequency data for the occurrence and distribution of non-extraposition in the ICE-GB corpus. These quantitative ªndings already contain some indications as to preferred patterns of use, but also relate to the subsequent functional discussion (cf. Section 4), which attempts to provide communicative-functional explanations for the distributional patterns found in the corpus.
3.1 Overall frequencies As was indicated in Section 1, in terms of its overall frequency, non-extraposition is by far outnumbered by its extraposed counterpart (see also Figure 1 below). With a total of 217 occurrences as opposed to 1,701 for it-extraposition, non-extraposition constitutes as little as 11.3 per cent of the overall number of subject clauses investigated in this study. Similar ratios are found for spoken and written texts (9.8% and 12.4% respectively). From a statistical point of view, therefore, the prototypical structure is it-extraposition, while non-extraposition, though often considered the structural prototype (cf. e.g. Huddleston 1984: 451), represents the marked variant. 2000
1701
1500 971
730
1000 500 0
79
138
Spoken
Written
non-extraposition
217 Total it-extraposition
Figure 1. Frequency of non-extraposition compared with subject it-extraposition
Using non-extraposition in spoken and written texts 223
731
800
665
600 400 200 0
12
41 56
34
that-cl.
wh-cl.
to-inf. non-extraposed
107 2 for/to-inf.
128 48 -ing cl.
0
55
other
extraposed
Figure 2. Frequency of non-extraposition and subject it-extraposition according to syntactic type of the subordinate clause
However, not all syntactic subclasses yield the same results, as can be seen from Figure 2: -ing clauses, in fact, oppose the general trend. As the more noun-like constituents (cf. also discussion of length in Section 4.1), they clearly prefer nonextraposition (128 vs. 48 instances), which therefore has to be considered their unmarked position. Wh-clauses show no clear preference for either position (41 instances non-extraposed, 56 extraposed). That-clauses in non-extraposition, on the other hand, are conspicuously rare (12 vs. 731 instances), as are for/to-inªnitives (2 vs. 107 instances). Generally speaking, non-extraposition represents a highly marked variant and it is therefore interesting to investigate the factors responsible for its use, which will be done in Section 4 below.
3.2 Spoken and written mode If we look at the distribution of non-extraposition in spoken and written texts,4 we notice that it is more likely to be encountered in the written category, which contains almost twice as many instances as the spoken (63.6% vs. 36.4%; 138 vs. 79 instances). This preference for the written mode is at least partly attributable to the extra processing eŸort required by clausal subjects in non-extraposed, i.e. structure-initial, position, which runs counter to the ‘end-weight principle’ or ‘light subject constraint’ (Chafe 1991), principles that apply more rigorously in unplanned speech (cf. also discussion in Sections 4.1 and 3.3 below). By comparison, it-extraposition is also more frequent in the written mode, however, less markedly so, with the ratio being 57.1 per cent vs. 42.9 per cent. From this predominance of written non-extrapositions over spoken ones, however, we cannot deduce a possible dependency between the two structural types, i.e. an in¶uence of the independent variable ‘mode’ on the choice between extraposition and non-extraposition. A test of statistical signiªcance (chi-square)
224 Gunther Kaltenböck
shows that a preference for non-extraposition is not signiªcantly aŸected by the spoken-written divide (χ2 = 2.969 [cf. footnote 5] crit. 3.841).5 The decisive factor for the choice of non-extraposition must therefore be sought elsewhere, viz. in its functional properties. The two constructions also diŸer in the distribution pattern of their syntactic subtypes, as is illustrated in Figure 3. While it-extraposition shows a fairly heterogeneous picture with some syntactic types leaning more towards the written mode (viz. that-clause 59%, inªnitive 60.2%) and others occurring more often in spoken texts (viz. wh-clause 55.4%, -ing clause 58.3%), non-extraposition prefers the written mode throughout, i.e. irrespective of syntactic type. The degree of preference, however, varies. Non-extraposed wh-clauses are somewhat unusual in that they show a fairly balanced distribution between the two modes (48.8% spoken, 51.2% written). With all other types of non-extraposed subject clauses written instances clearly predominate: 69.4 per cent for inªnitives (including for/to-inªnitives), 62.5 per cent for -ing clauses, and a striking 100 per cent for that-clauses, which are not attested in spoken texts at all.6 36.4
non-ex. total
Written %
63.6
0 100
Spoken %
48.8 51.2
wh-cl. 30.6
69.4 37.5
-ing cl.
62.5
50
0
100
42.9
extrapos. total
41
wh-cl.
44.6
57.1
Written %
59
Spoken %
55.4
39.8 60.2
-ing cl.
41.7 0
20
40
58.3 60
80
100
Figure 3. Distribution of non-extraposition and it-extraposition in spoken and written texts
Using non-extraposition in spoken and written texts 225
3.3 Text type As regards the distribution of non-extraposition in diŸerent text categories, it has been claimed (e.g. Collins 1994: 14–15) that the pressure to extrapose increases with the degree of informality. Accordingly, non-extraposition can be expected to occur more frequently in formal texts. While formality is a somewhat di¹cult concept to pin down, the frequency ªndings in the corpus, presented in Figure 4, suggest that there is indeed a formality-informality cline for non-extraposition. Figure 4 shows that the number of non-extrapositions increases from the spoken (ªrst three bars on the left) to the written categories as well as within ‘spoken’ and ‘written’ from the more informal to the more formal texts. The three spoken text types in ICE-GB reveal a slight but steady increase of non-extrapositions from the least formal category ‘Private dialogue’ via ‘Unscripted monologue’ to the more formal category ‘Public dialogue’. In the written texts there is considerably more variation. Interestingly, the lowest ªgures are found in ‘Creative writing’ (i.e. novels and stories), which can be explained by the frequent use of direct and indirect speech (i.e. recreation of spoken language) in this text type. The more informal or speech-like categories ‘Correspondence’ and ‘Scripted monologue’ yield slightly fewer non-extrapositions than the more formal categories ‘Non-professional writing’, ‘Non-academic writing’, ‘Academic writing’ and ‘Instructional writing’. Special cases are ‘Reportage’ and ‘Persuasive writing’ with Non-extrapositions (per 10,000 words) 6
5,5 4,8
5
2,9
3
3
Instructional writing
2,8
3
Academic writing
4 2,3 1,7
2
1,9
1,2
1,8 1
1
Persuasive writing
Reportage
Non-academic writing
Non-professional writing
Scripted monologue
Correspondence
Creative writing
Public dialogue
Unscripted monologue
Private dialogue
0
Figure 4. Frequency of non-extraposition in the ICE-GB text categories
226 Gunther Kaltenböck
unusually high occurrences of 4.8 and 5.5 respectively. The category ‘Persuasive’, in other words, contains 5.5 times more non-extraposed subject clauses than ‘Creative writing’. This high frequency is only in part attributable to the factor of formality, though. For an explanation we also have to look at the syntactic form of subject clause involved: in both categories (‘Reportage’ and ‘Persuasive writing’) nonextraposition yields more than twice as many -ing clauses (3.0 and 4.0 occurrences per 10,000 words) than in the average written text (1.3 occurrences). With -ing clauses preferring subject position the high number of non-extrapositions is not really surprising. A further reason for the frequent use of non-extrapositions in these texts is provided by a speciªc discourse function of this constructional variant, which will be discussed in Section 4.2.3. When considering the in¶uence of formality of text types on the use of nonextraposition it is thus necessary again to diŸerentiate between the syntactic types of subject clause involved. The susceptibility to diŸerent degrees of formality is most noticeable with non-extraposed to-inªnitives, which show considerable variation especially within the spoken mode: all instances occur in the more formal categories of ‘Monologue’ or ‘Public dialogue’, the informal genre of ‘Private dialogue’ containing none. Within the written mode they are most frequent in the text categories ‘Academic writing’ and ‘Persuasive writing’, while the lowest count is found in ‘Correspondence’ and ‘Scripted monologue’ (cf. similar ªndings for academic prose in Biber et al. (1999: 724)). This clearly marks non-extraposed toinªnitives as a construction of formal texts. With non-extraposed wh-clauses the in¶uence of formality is somewhat less obvious; that-clauses are too small in number to warrant a valid conclusion. Non-extraposed -ing clauses, on the other hand, show little variation in the spoken register, and their frequency in the written text categories does not seem to be directly linked to the formality-informality continuum but rather associated with a particular style of writing, which is typical of newspaper articles and press editorials. This in turn relates to the speciªc communicative eŸect that can be achieved by non-extraposition. For an explanation of its distribution in the diŸerent text types it is therefore necessary to examine its functional potential, which will be done in the following section.
4.
Functional analysis
Given the structural and semantic interchangeability7 of non-extraposition and itextraposition, which is indeed the deªning criterion for this syntactic paradigm (cf. discussion in Section 2), there remains the question of why the non-extraposed variant is chosen less frequently — especially in spoken texts — so that it constitutes a statistically marked option. Conversely, it is also interesting to investigate
Using non-extraposition in spoken and written texts 227
why non-extraposition, being so restricted in use, is chosen at all. The answer to these questions must be sought in the functional characteristics of this constructional type, more precisely the distribution of weight and information within the construction, which will be discussed in Sections 4.1 and 4.2 respectively.
4.1 Weight of subject and matrix clause The main reason for the infrequent occurrence of non-extraposition in spoken texts as well as its overall low occurrence compared to its extraposed counterpart lies, no doubt, in the extra processing eŸort required for structure-initial clausal subjects. In pre-predicate position they run counter to the ‘end-weight principle’ (cf. also Chafe’s 1991 ‘light subject constraint’), which is much more extreme in spoken language (cf. e.g. Biber et al. 1999: 1067). Indeed, the notion of ‘weight’ is often evoked as the decisive factor for the choice between extraposition and non-extraposition (cf. e.g. Jespersen 1949, VII: 57; Langacker 1974: 653; Huddleston 1984: 452). An analysis of actual corpus data, however, is only incorporated in the studies by Erdmann (1988) and Collins (1994), both of which are considerably restricted in that the former takes into account exclusively written (non)-extrapositions with an adjectival matrix predicate and a to-inªnitive or that-clause, and the latter is based on a small-scale corpus containing only 20 instances of non-extraposition. Moreover, Erdmann’s study is limited in so far as it measures ‘weight’ in terms of syntactic complexity. While syntactic complexity undoubtedly contributes to the heaviness of a constituent it is an inherently problematic concept as it does not easily allow the comparison of diŸerent degrees of complexity needed for establishing the relative weight of two constituents (Erdmann therefore resorts to a somewhat crude binary light-heavy classiªcation). It is, however, precisely the weight of subject and matrix clause in relation to each other that is of interest in the case of (non)-extraposition with its two-part structure (for further criticism of Erdmann’s approach cf. Collins 1994: 16). The present study, therefore, follows Collins (1994) in adopting the practice of word-count, which facilitates comparison of the ‘weight’ of the two constituents in question. The results (given in Table 1) show that the extraposed subject clause is on average three times longer than its matrix clause (3.5 vs. 11.7 words). By comparison, the non-extraposed subject clause is not necessarily shorter than its matrix predicate, as might have been expected. In fact, the picture presented by nonextraposition is a very balanced one with matrix and subject clause being, on average, equal in length (viz. 7.8 words each). This distribution is interesting as it would in theory just as well allow the reverse order (i.e. extraposition). If we analyse the relative length of matrix and subject clause in more detail, we ªnd that the majority of non-extrapositions (viz. 46.3%) have a subject clause which is longer
228 Gunther Kaltenböck
Table 1. Average length (in words) of matrix and subject clause in non-extraposition and it-extraposition Non-extraposition Subject clause Matrix clause Spoken Written Total
6.6 8.9 7.8
8.0 7.6 7.8
It-extraposition Matrix clause Subject clause It It It
3.7 3.2 3.5
10.7 12.7 11.7
than its matrix clause, in 12.5 per cent of the cases they are of equal length, and only 41.2 per cent have a longer matrix clause. It needs to be pointed out, however, that the matrix clause of non-extraposition is at least twice as long (7.8 vs. 3.5) as that of it-extraposition, which supports the end-weight principle. Thus, it seems that as long as the matrix predicate in non-extraposition satisªes a certain minimum length (in absolute terms) the principle of relative weight may easily be overridden. As is to be expected, there are noticeable diŸerences in the weight distribution of spoken and written non-extrapositions: spoken ones display a ratio of 6.6 (Subject clause): 8.0 (Matrix clause), while written ones have a ratio of 8.9 (Subject clause): 7.6 (Matrix clause). In other words, the end-weight principle seems to have more ‘weight’ in spoken instances of non-extraposition. This is, of course, not really surprising, as questions of processing eŸort are more immediately relevant in spoken language, where placing long elements towards the end of the clause considerably eases comprehension (and indeed production) since the listener/ speaker does not have to retain complex information from earlier in the clause while processing/producing the remainder. Apart from the diŸerences in spoken and written texts, non-extraposition also displays variation for its diŸerent syntactic subtypes. Table 2 illustrates that only non-extraposed -ing clauses are, on average, shorter than their matrix predicates, which explains their frequent occurrence in pre-predicate position (cf. Section 3.1). All other syntactic types (to-inªnitive,8 wh-clause, that-clause) show a rather unusual heavy-light pattern in both modes.
Table 2. Average length (in words) of subject clause and matrix clause in nonextraposition according to syntactic type of the subject clause Spoken Subject clause Matrix clause to-inªnitive that-clause wh-clause -ing clause
10.9 0 7.5 5.9
> = <
7.1 0 7.5 7.8
Written Subject clause Matrix clause 10.5 11.4 12.9 6.9
> > > <
7.5 9.5 6.4 7.5
Using non-extraposition in spoken and written texts 229
The preponderance of unusually long subject clauses in construction-initial position, which runs counter to the ‘light subject constraint’ (Chafe 1991), seems to suggest that the principle of balanced weight distribution is easily ‘outweighed’, as it were, by another important factor, viz. that of achieving a particular communicative eŸect. This communicative function of non-extraposition is associated with the question of information packaging in the construction, which will be discussed in some detail in the following.
4.2 Information packaging The information status of the subordinate clause has variously been evoked as a factor in¶uencing the choice between the extraposed and non-extraposed variant. Huddleston (1984: 453–454), for instance, speculates that “a context where the content of an embedded ªnite clause or inªnitival is given may favour the unmarked construction [= non-extraposition]”. Erdmann (1987: 49) goes one step further and describes non-extraposition as an anaphoric construction whose clausal subject always refers back to the previous linguistic context, a view also expressed by Collins (1994: 21), who maintains that non-extraposed constructions “do not occur unless the clausal subject contains at least some given information”, and Miller (2001: 688–689, 693–695). None of these studies, however, attempts an investigation of the information structure of (non)-extraposition on a larger scale. Given the amount of terminological confusion in this ªeld, it is necessary, ªrst of all, to determine what is actually meant by the terms ‘given’ and ‘new’ information. Following Prince (1992) and Mackenzie & Keizer (1990) it is possible to establish essentially two diŸerent views of information status: given/new in the hearer’s/speaker’s head, i.e. the speaker’s beliefs about the hearer’s beliefs, or ‘assumed familiarity’ (cf. also Chafe’s e.g. 1976, 1987 notion of consciousness), and given/new in the discourse model, i.e. whether a state-of-aŸairs has already been evoked in the previous discourse or not. While a deªnition of givenness in terms of speaker assumptions may well capture more accurately the psychological reality of the concept, it is highly unoperational and impossible to apply to corpus data as we can only speculate on what must have gone on in the speaker’s head. For the present purpose, therefore, I have adopted the concept of retrievability, as used for instance by Firbas (e.g. 1992) and Geluykens (e.g. 1991), according to which, information is ‘given’ if it is retrievable (recoverable) from the preceding (verbal) cotext or (situational) context and ‘new’ if it is not thus derivable. The retrievability span allowed for in this study follows Givón’s (1983: 13) upper limit of 20 clauses to the left, but may extend beyond it, e.g. in conversations consisting of short turns, or texts where a particular state-of-aŸairs represents a prominent discourse topic. Retrievability, of course, is a relative concept with some items
230 Gunther Kaltenböck
being more immediately retrievable from the preceding co(n)text than others. To account for the various degrees of givenness/newness, the present study makes use of Prince’s (1981a) taxonomic model of ‘assumed familiarity’ and adapts it to the concept of retrievability, i.e. a discourse-based model of ‘discourse familiarity’. It distinguishes between two types of new (irretrievable) information, viz. ‘brandnew’ and ‘new-anchored’ with the latter representing a new state-of-aŸairs that is in some way linked to the previous discourse (e.g. by the use of anaphora or word repetition), and three types of givenness, viz. ‘inferrable’, ‘textually evoked’ and ‘situationally evoked’. Whereas the latter two categories refer to states-of-aŸairs that have been explicitly evoked either by the previous text or extralinguistic situation, the category of ‘inferrables’ applies where the clausal subject expresses a state-of-aŸairs that can be inferred (retrieved indirectly) from some ‘trigger entity’ in the preceding context. The following examples illustrate the categories ‘textually evoked’ (5), ‘inferrable’ (6) and ‘new-anchored’ (7).9 (5) a.
Faced with economic recession, a radical conservative government would have cut spending. This conventional conservative government is to increase it by 2.2 per cent in real terms next year […] The economy has shrunk between the ªrst half and second half of this year, and it will stagnate over the ªrst six months of next year. In these circumstances, t o cut public spending would be hideously di¹cult (W2C-008-77) b. I’m very proud of being British as well and I admit that tears come to my eyes when I watch something like the recent services commemorating the Battle of Britain But for me being patriotic has nothing to do with our monarchy (S2B-032-6)
(6) a.
And before President Gorbachev had barely set foot back on Soviet soil the Sojuz group had passed a vote of no conªdence in him a decision to gather a petition to call an emergency session of the Congress of People’s Deputies the only body that has the power to oust him as President To collect signatures for that petition will take time (S2B-040-59) b. The temperature drops to minus sixty […] Every night they pitchcamp precariously in the snow-covered wilderness sleeping ªtfully on dangerous shifting pack-ice that shrieks as it moves There’s the ever-present fear that the ice will shear oŸ dragging you down into the open chasm beneath you Getting up each morning in sub-zero temperatures after a few hours’ broken sleep calls for collossal will-power (S2B-024-95)
(7) a.
And it certainly would be peculiar if all the myths about it [=the class system] were actually true but of course they’re not For example in
Using non-extraposition in spoken and written texts 231
nineteen eighty-four according to a Gallup poll seventy percent of the British population claimed to be working class claimed not only that they were working class but their parents were working class According to sociologists the percentage was at least twenty percent less That Labour supporters and those on the left of Labour should believe this myth is obvious (S2B-035-86) b. Worse still, they [=the United States] feared that once in India’s hands the technology would leak out to the Soviet Union. However, at this meeting the American president agreed that India should have the computer. Whether this was an indication that India was henceforth to be taken off the ‘suspicion’ list or simply a sweetener in the hope that the United States would be well placed when India made up its mind on its defence shopping list, is a matter for speculation (W2B-011-57) An analysis of the information status of the extraposed and non-extraposed clausal subjects in the corpus yields the following results (Figures 5 and 6).
100
19,8
7,6 26,8
80 60 80,2
40
92,4 73,2
20 0 Total
Spoken
'Given' subject clause %
Written New' subject clause %
Figure 5. Information status of the non-extraposed subject clause 100 80 60
71,5
56,1 83,2
40 20
28,5
43,9 16,8
0 Total 'Given' subject clause %
Spoken
Written New' subject clause %
Figure 6. Information status of the extraposed subject clause
232 Gunther Kaltenböck
We can see that there is a clear diŸerence between the preferred information status of the subject clause in extraposition and non-extraposition. In non-extraposition the overwhelming majority (80.2%) of subject clauses convey ‘given’ information, i.e. states-of-aŸairs which are in some sense retrievable from the preceding co(n)text. This preference for given information applies to all syntactic types of nonextraposition but is even stronger in spoken texts, where given subject clauses account for 92.4 per cent as opposed to 73.2 per cent in written texts. By comparison, it-extrapositions show the reverse pattern with a clear preference (71.5%) for ‘new’ subject clauses although much less markedly so in spoken texts (56.1%). Having established the information structure of non-extraposed clauses we can now turn to a discussion of the discourse function of non-extraposition, which will have to take into account both possible types, i.e. Given Subject Non-extraposition and New Subject Non-extraposition, as well as questions of sequential ordering. It is thus possible to identify three main functional characteristics of non-extraposition, which will be discussed in the following. The ªrst of these (cf. Section 4.2.1) applies to all non-extrapositions, irrespective of the information status of the subject clause; the other two features (cf. Sections 4.2.2 and 4.2.3) apply to Given Subject Non-extraposition and New Subject Non-extraposition respectively.
4.2.1 Focussing on speaker comment Like it-extraposition, non-extraposition expresses some comment (in the form of the matrix predicate) on some topic conveyed by the subordinate clause (cf. Lutz 1981: 40–41 on assigning a topic-comment structure to extraposition/non-extraposition). Where they diŸer is in the sequential ordering of the two: non-extraposition presents the matrix predicate in construction-ªnal position. This end-focus position of the predicate and its typically high communicative value, at least relative to its subject clause, which — as we have seen above — tends to represent given information, has the eŸect of adding extra emphasis to the comment of the speaker or writer. Compare the following examples: (8) a.
To be innovative in how you handle your accounts can be equally important (S2A-037-117) b. It can be equally important to be innovative in how you handle your accounts
Compared to the extraposed version (8b), the non-extraposition (8a) takes the focus oŸ the subject clause, making the matrix predicate, i.e. the speaker’s comment (evaluation), the main focus of the message. It-extraposition, apart from moving the speaker comment out of focus position, also has the eŸect of ‘depersonalising’ the comment by using impersonal it.
Using non-extraposition in spoken and written texts 233
This allows the speaker to express a personal comment towards some state-ofaŸairs, but present it as if it were the generally accepted view rather than the personal judgement of the speaker (cf. Biber et al.’ s 1999: 977 “implicit attribution of stance to the speaker/writer”). Such ‘masking’ of a personal evaluation has the pragmatic eŸect of not only reducing speaker responsibility by deferring it to some external authority and therefore making it less contestable, but may also add a ¶avour of objectivity and authority to the utterance. This particular function of itextraposition is often exploited in texts with a fairly ‘matter-of-fact’ or authoritative style (e.g. scientiªc texts); cf. e.g. It is well-known/believed/generally agreed that… With non-extraposition putting considerable emphasis on the speaker/writer comment (i.e. the matrix predicate) it is not surprising to ªnd a substantial number of non-extrapositions in text types with a tendency towards evaluation and expressing writer stance, such as newspaper editorials, i.e. our category of ‘Persuasive writing’, and to some extent ‘Reportage’ (cf. Section 3.3). In individual texts, non-extraposition is often used when the information of the matrix predicate, i.e. the climax of the message, is further elaborated or speciªed in the immediately following text. Compare the example in (9), where the corresponding it-extraposition (It is not yet certain whether…) would disrupt the smooth transition from the matrix predicate to the subsequent elaboration, viz. that, despite the uncertainty, there is no objection to the state-of-aŸairs in question. (9) Whether or not peptide receptors are also able to gate ion channels directly analogous to the nicotinic acetylcholine receptor-channel is not yet certain, although there is no objection to it in principle (W2A-027-32) However, it is also common for the matrix predicate of non-extraposition to represent the ªnal focal point of a paragraph or turn, providing the concluding comment on a particular discourse topic, which may be summarised in the subject clause. The ªnal matrix predicate shifts the discussion to a personal level expressing some ªnal evaluation, which signals that the speaker/writer regards the topic of discussion as exhausted. Compare example (10) (where
indicates the beginning of a paragraph,
its end): (15) a. The majority of the applicants are known to me and I am therefore aware that we possess the highest qualiªcations for such a post, i.e. […] All this information was clearly available in writing, together with
236 Gunther Kaltenböck
references, rendering an hour’s interview almost super¶uous. Why, then, was it not possible to appoint a team of three people from these candidates? Each of us would have happily accepted the Panel’s decision, but to appoint no-one from such highly qualiªed teachers was an insult to us all.
The decision to investigate an accident almost always implies that a visit by an inspector will be made to carry out an on-site investigation. The inspector making the visit could be anything from a trainee to a principal inspector. (Indeed, the accident may be selected for investigation largely because of the value to trainers of investigating a range of accidents.) A handful of less serious accident investigations fall by the wayside before a visit is made because of the time that has elapsed. For all the inspectorates we studied, ‘accident investigation’ is a class of visit in itself. However, an investigation is often combined with other activities. Thus, an inspector may use the same visit to check on other matters already known about, or to conduct a full basic inspection. About 5 per cent of Factory Inspectorate
Using non-extraposition in spoken and written texts 237
investigations bring the inspector to premises not previously registered with the HSE. Investigating the accident may then be secondary to impressing on the employer the requirement to register, and ªnding out more about the previously unknown and uninspected establishment that has been brought to light.
The more teams practice and rehearse their set moves, the more accurate and well-coordinated their performance will become […] There is a further beneªt from rehearsing moves. It helps players develop their concentration. It is not enough for a player to know what he should be doing. He must concentrate in order to do it eŸectively and to do it whenever it is required. Players who lose concentration tend to drift out of the game.
Rehearsing pre-planned moves helps players to be alert and watchful (W2D-015-107) b. A: Tim Renton as the Minister you’re cast as the villain of the piece in this the prince of darkness Do you accept that that you are the villain And do you accept that there is a crisis E: I think being Prince of Darkness is actually an attractive title isn’t it (S1B-022-66)
4.2.3 Presenting a topic as given As pointed out in Section 4.2, a small number of non-extraposed subject clauses (19.8%, 43 instances) convey states-of-aŸairs which are ‘new’, i.e. irretrievable from the preceding context. Even though the large majority of these occur in written texts, where ease of processing is arguably less immediately relevant than in spoken texts, this way of presenting new information (viz. in construction-initial position) is still unusual. Not only does it violate the given-before-new principle, but also the principle of end-weight (light subject constraint): the average length for this type of subject clause in written texts is 11.5 words, as opposed to only 7.4 words for its matrix clause. Clearly, the expected position for these long and new clauses would be in extraposition, which is indeed where most of the irretrievable
238 Gunther Kaltenböck
subject clauses end up: 83.2 per cent of all extraposed clauses in written texts convey new information. Why irretrievable subject clauses still occur in non-extraposed position can be explained by looking at their communicative function. Structure-initially, they take prototypical subject/topic position, which has come to be closely associated with givenness (cf. e.g. Horn 1986; Prince 1986: 209, 1981b). Moreover, they also represent the ‘topic’ (in terms of what the construction is about; cf. Lutz 1981: 40–41) about which some ‘comment’ (the matrix predicate) is being made and they are syntactically backgrounded (cf. Mackenzie 1984, Sadock 1984 on the pragmatic eŸect of syntactic subordination). Altogether this results in an eŸect of downgrading and presupposition. Thus, in spite of its actual newness in terms of retrievability from the preceding context, the state-of-aŸairs in the subject clause is presented as if it were given information. Compare, for instance, example (18), the opening line of a party political broadcast where Building a top-class economy, although coming ‘out of the blue’, is presented as the generally known aim of Labour policy, as a given fact that we are all familiar with. (18) ## Building a top-class economy a world-class economy will take time eŸort resources and consistency of purpose and policy. So much is evident (S2B-030-107) This presupposition eŸect is even more obvious with that-clauses and their explicit use of a subordinator, as in (19a) and (19b): (19) a.
[Architecture, Philosophy and the Public World] Architecture being the paradigm of public art, its philosophical examination is an exercise in social aesthetics.
That these matters are currently a topic of considerable public and professional discussion is due in large part to the several widely reported occasions on which the Prince of Wales has castigated modern architects for producing ‘gross inhuman monstrosities’. (W2A-005-11) b. [Lecture: Who owns my cells?] Implicit in the court’s reasoning is the assumption that ownership is absolute or it’s not ownership That this is wrong hardly I think needs demonstration (S2B-046-87)
In this function non-extraposition is useful for contexts where a writer wants to introduce information which may be known to some readers but not necessarily to all, as a diplomatic way of indicating ‘this is something you should know’, ‘I am taking this to be general knowledge’ (cf. example (19a)), or as a rhetorical device for deliberately ‘masking’ new information as ‘given’ and presenting it as a generally
Using non-extraposition in spoken and written texts 239
known fact (cf. examples (19b) and (18)). Not surprisingly, therefore, new subject clauses in non-extraposed position are most frequently to be found in the text categories ‘Academic writing’, ‘Reportage’, and ‘Persuasive writing’ (with frequencies of 1.1, 1.3 and 1.5 occurrences per 10,000 words respectively as opposed to an average frequency of 0.4 in the total corpus), which can provide some explanation for the overall high frequency of non-extraposition in these text types, especially ‘Reportage’ and ‘Persuasive writing’ (cf. Section 3.3).
5.
Conclusion
Despite their structural relationship (which typically allows ‘transformation’ into the other variant) and semantic synonymy (i.e. their propositions are truth-conditionally equivalent and both represent predicational constructions expressing some ‘comment’ on a clausal ‘topic’), it-extraposition and non-extraposition show a very diŸerent behaviour in their actual use in spoken and written texts. Indeed, from a functional perspective overlap between the two constructional types is minimal and the extraposition paradigm has little more than abstract structural value. Each type is tied to a very speciªc context, which does not normally allow free ‘swapping’. Non-extraposition, as the statistically marked variant of the two, displays a number of very speciªc discourse functions, which are diŸerent from those of its extraposed counterpart. Thus, non-extraposed subject clauses, despite their length, resist the general tendency of being shifted into construction-ªnal position in order to allow the speaker/writer to focus on the matrix predicate and, in the majority of cases, to achieve a more balanced, i.e. given-before-new, distribution of information within the construction, which in turn establishes a strong cohesive link with the preceding context. In some cases, non-extraposed clauses may also serve the rhetorical purpose of introducing a new topic into the discourse, while presenting it as if it were generally known. Establishing the communicative properties of nonextraposition has enabled us not only to cast some light on the use of this construction in a text, but also to provide some explanations for its distribution across diŸerent text types as discussed in Section 3.
Notes 1. For a different view of anticipatory it cf. Kaltenböck (2003). 2. Notable exceptions are Erdmann (1987: 39–54), Mair (1990: 34–39), Collins (1994: 15– 22) and Miller (2001), who include brief discussions of communicative aspects. All of these studies are, however, limited either by the corpus material used or their scope of interest (e.g. only infinitival clauses in the case of Mair 1990).
240 Gunther Kaltenböck
3. Classiªcation of texts in ICE-GB follows the delivery mode and groups 50 texts of scripted speech (S2B, ‘written to be spoken’) with the spoken texts. In terms of production, however, this category diŸers from other orally delivered texts in that it is not produced ‘online’, as it were, but with a degree of premeditation and planning typical of written texts. Following the arrangement of other corpora (e.g. the Survey of English Usage Corpus) ‘scripted speech’ is therefore grouped here with the written text category. This decision is also supported by frequency counts, which show that the occurrence of non-extraposition in scripted speech is in line with the average result for written texts. 4. The written and the spoken category comprise 250 texts each; cf. also note 3. 5. Similarly, the hypothesis that the entire grammatical choice (non-extraposition or itextraposition, but not specifying which one) is significantly affected by the spoken-written category could also not be confirmed (χ2 = 3.467 < crit. 3.841). 6. The figure for extraposed that-clauses includes cases of that-omission (zero that). 7. Some restrictions do apply to a small class of matrix predicates, e.g. it seems / appears / turns out, and agentless passives, e.g. it is believed / assumed / suggested, which block nonextrapostion (cf. *That he did it seems, *That she went there is believed), but are typically classified as instances of it-extraposition (cf. e.g. Quirk et al. 1985: 1392). They are, however, marginal cases of it-extraposition and therefore disregarded in the present discussion. 8. The category of for/to-infinitives has been omitted here since there are only two instances in the corpus, one written and one spoken, with the latter having an incomplete matrix clause. 9. For further examples and a more detailed discussion of the diŸerent types of information status cf. Kaltenböck (2001).
References Biber, Douglas, Johansson, Stig, Leech, GeoŸrey, Conrad, Susan and Finegan, Edward 1999 Longman Grammar of Spoken and Written English. London: Longman. Chafe, Wallace L. 1976 “Givenness, contrastiveness, deªniteness, subjects, topics, and point of view”. In Subject and Topic, C. N. Li (ed.), 22–55. New York: Academic Press. 1987 “Cognitive constraints on information ¶ow”. In Coherence and Grounding in Discourse, R. S. Tomlin (ed.), 21–51. Amsterdam and Philadelphia: John Benjamins. 1991 “Grammatical subjects in speaking and writing”. Text 11: 45–72. Collins, Peter 1994 “Extraposition in English”. Functions of Language 1: 7–24. Emonds, Joseph 1972 “A reformulation of certain syntactic transformations”. In Goals of Linguistic Theory, S. Peters (ed.), 21–62. Englewood CliŸs, NJ: Prentice-Hall. 1976 A Transformational Approach to English Syntax: Root, Structure-Preserving, and Local Transformations. New York: Academic Press.
Using non-extraposition in spoken and written texts 241
Erdmann, Peter 1987 It-Sätze im Englischen. Heidelberg: Carl Winter. 1988 “On the principle of ‘weight’ in English”. In On Language: Rhetorica, Phonologica, Syntactica: A Festschrift for Robert P. Stockwell from his Friends and Colleagues, C. Duncan-Rose and T. Vennemann (eds), 325–339. London: Routledge. Firbas, Jan 1992 Functional Sentence Perspective in Written and Spoken Communication. Cambridge: Cambridge University Press. Geluykens, Ronald 1991 “Information ¶ow in English conversation: A new approach to the givennew distinction”. In Functional and Systemic Linguistics: Approaches and Uses, E. Ventola (ed.), 141–167. Berlin: Mouton de Gruyter. Givón, Talmy 1983 “Topic continuity in discourse: An introduction”. In Topic Continuity in Discourse: A Quantitative Cross-Language Study, T. Givón (ed.), 1–41. Amsterdam and Philadelphia: Benjamins. 1995 Functionalism and Grammar. Amsterdam and Philadelphia: Benjamins. Halliday, M. A. K. 1994 Introduction to Functional Grammar (2nd ed.). London: Edward Arnold. Horn, Laurence R. 1986 “Presupposition, theme, and variations”. In Papers from the Parasession on Pragmatics and Grammatical Theory at the 22nd Regional Meeting, A. M. Farley, P. T. Farley and K.-E. McCullough (eds), 168–192. Chicago: Chicago Linguistic Society. Huddleston, Rodney 1971 The Sentence in Written English. Cambridge: Cambridge University Press. 1984 Introduction to the Grammar of English. Cambridge: Cambridge University Press. Jespersen, Otto 1949 A Modern English Grammar on Historical Principles. Part VII. London: Allen & Unwin. Kaltenböck, Gunther 2000 “It-extraposition and non-extraposition in English discourse”. In Corpus Linguistics and Linguistic Theory: Papers from the 20th International Conference on English Language Research on Computerized Corpora (ICAME 20), C. Mair and M. Hundt (eds), 157–175. Amsterdam: Rodopi. 2003 “On the syntactic and semantic status of anticipatory it”. English Language and Linguistics 7 (2): 235–255. Langacker, Ronald W. 1974 “Movement rules in functional perspective”. Language 50: 630–664. Lutz, Luise 1981 Zum Thema ‘Thema’. Einführung in die Thema-Rhema Theorie. Hamburg: Hamburger Buchagentur.
242 Gunther Kaltenböck
Mackenzie, J. Lachlan 1984 “Communicative functions of subordination”. In English Language Research: The Dutch Contribution I, J. L. Mackenzie and H. Wekker (eds), 67– 84. Amsterdam: Free University Press. Mackenzie, J. Lachlan and M. Evelien Keizer 1990 “On assigning pragmatic functions in English”. Working Papers in Functional Grammar 38 (Pragmatic functions: the view from the V. U.): 24–56. Mair, Christian 1990 Inªnitival Complement Clauses: A Study of Syntax in Discourse. Cambridge: Cambridge University Press. Miller, Philip H. 2001 “Discourse constraints on (non)extraposition from subject in English”. Linguistics 39 (4): 683–701. Prince, Ellen F. 1981a “Toward a taxonomy of given-new information”. In Radical Pragmatics, Peter Cole (ed.), 223–255. New York: Academic Press. 1981b “Topicalization, Focus-Movement, and Yiddish-Movement: A pragmatic diŸerentiation”. In Proceedings of the 7th Annual Meeting of the Berkeley Linguistics Society, D. Alford et al. (eds), 249–264. Berkeley: Linguistics Society. 1986 “On the syntactic marking of presupposed open propositions”. In Papers from the Parasession on Pragmatics and Grammatical Theory, 22nd Regional Meeting, A. Farley, P. Farley and K.-E. McCullough (eds), 208–222. Chicago: Chicago Linguistic Society. 1992 “The ZPG letter: subjects, deªniteness, and information-status”. In Discourse Description: Diverse Analyses of a Fund-Raising Text, W. C. Mann and S. A. Thompson (eds), 295–325. Amsterdam and Philadelphia: John Benjamins. Quirk, Randolph, Greenbaum, Sidney , Leech, GeoŸrey and Svartvik, Jan 1985 A Comprehensive Grammar of the English Language. London: Longman. Rosenbaum, Peter S. 1967 The Grammar of English Predicate Complement Constructions. Cambridge, Mass.: M. I. T. Press. Sadock, Jerrold M. 1984 “The pragmatics of subordination”. In Sentential Complementation, W. de Geest and Y. Putseys (eds), 205–213. Dordrecht: Foris. SchegloŸ, Emanuel A., JeŸerson, Gail and Sacks, Harvey 1977 “The preference for self-correction in the organization of repair in conversation”. Language 53: 361–382.
Part IV
Metaphor and text
English metaphors and their translation The importance of context* Kay Wikberg University of Oslo
1.
Introduction
To begin with, we shall have a look at a metaphor found in Stephen King’s novel Cujo, which opens with a description of a mentally deranged policeman who has killed six people. After his death the policeman is still haunting the town. A fouryear-old boy sees him appearing as a monster in a cupboard at night making that hissing noise. The metaphor describes the monster’s way of breathing. This example from the multilingual Oslo corpus demonstrates that the original caused some di¹culty in translation: (1) his breath a thin winter-whistle in his throat (SK1.1.s26) N Pusten lød som en hes ¶øyte i halsen hans S untranslated F hengitys takellellen G er wagte kaum zu atmen (“he hardly dared to breathe”) (SK = Steven King; 1.1.s26 = Chapter 1, paragraph 1, sentence number 26; N = Norwegian, S = Swedish, F = Finnish, G = German) His refers to the monster in the cupboard. A comparison of the translations shows that the Norwegian version is by far the best although it adds a simile (lød som en hes ¶øyte = ‘sounded like a hoarse ¶ute’), which is claimed to reduce metaphoricity. The Swedish translator gave up for no apparent reason, and the German translation went wide of the mark. The Finnish version has lost the metaphor, but uses a verb which at least refers to the monster’s way of breathing, as if ‘stammering’. The arrival of the monster and the use of metaphors to describe it are well heralded in the previous text. For example, the monster’s eyes have just been described as amber-glowing pits. The syntax is marked as well: “Low to the ground it
246 Kay Wikberg
was, with huge shoulders bulking above its head…” As regards the metaphorical expression, the Vehicle winter-whistle – – – –
is a unique compound, which would have ended up on a list of hapax legomena; has an unconventional reference (breath); is not separated by punctuation from the previous Noun Phrase, and therefore cannot be an apposition; can be preceded by an inserted be and can therefore be interpreted as a Subject Complement.
It has been argued that English is not particularly rich in sensory lexis, which might be some support for the use of an image like this. Another and more plausible justiªcation for the metaphor in (1) would be that by using this semantically dense ªgurative expression Stephen King wanted to say something which would have been hard to express by nonliteral means. To ªnd a metaphor like the one described here without ªrst reading the text would have been di¹cult. Innovative metaphors, which make up most of the data for this paper, cannot be detected by means of computer unless they contain distinctive markers, and most metaphors do not. One methodological question I would like to answer in this paper is then: How can corpus-based research help us to throw light on metaphor and the translation of metaphor? I shall try to give a partial answer to this question. Contrary to practice in most corpus studies I shall have little to say about quantities. The nature of metaphor is simply such that it invites much less to quantiªcation than to qualitative analysis. This is probably one reason why so little has been said about metaphor in connection with corpora. I would also like to stress that, in spite of the title, my intention in this context is not to say so much about purely contrastive ªndings since the main contrasts are known anyway. I will rather focus on current trends in metaphor research and their implications for the analysis of metaphors generally and in translation, and for the study of the functions of metaphors in discourse.
2.
The corpus
The Oslo multilingual corpus allows us to study a number of languages from a contrastive perspective. For example, starting out from English ªction texts, we can examine their translation into Norwegian, Swedish, Finnish, German, Dutch, and Portuguese. We can also use some of these other languages as the source language and investigate the translations of original texts in those languages into English. The Oslo corpus contains both ªction and non-ªction texts; for the present purpose I
English metaphors and their translation 247
was interested in ªction only.1 These are the sources of the extracts included in the subcorpus for this paper: Atwood, Margaret, Cat’s Eye (Canadian) Barnes, Julian, Talking It Over (British) Brink, André, The Wall of the Plague (South African) Dahl, Roald, Matilda (British; for children) Drabble, Margaret, The Middle Ground (British) Forsyth, Frederick, The Fourth Protocol (British; spy story) Francis, Dick, Straight (British; detective novel) Grafton, Sue, “D” is for Deadbeat (American; detective novel) James, P. D., Devices and Desires (British; detective novel) King, Stephen, Cujo (American) Townsend, Sue, The Queen and I (British) The authors of these books are all well established and represent both light and more serious subgenres. A question that arises in connection with these various types of ªction is what role metaphor might play in them. For example, can detective stories or thrillers be expected to diŸer from ‘ordinary’ ªction in their use of metaphor? If a book is written primarily to describe action and to entertain rather than for intellectual and/or aesthetic stimulus, I assume with Huntington Wright that there will be less need for ªgurative language: The style of a detective story must be direct, simple, smooth, and unencumbered. A “literary” style, replete with descriptive passages, metaphors, and word pictures, which might give viability and beauty to a novel of romance or adventure, would, in a detective yarn, produce sluggishness in the actional current by diverting the reader’s mind from the mere record of facts (which is what he is concerned with), and focussing it on irrelevant æsthetic appeals (W. Huntington Wright. 1927. The Great Detective Stories. http://gaslight.mtroyal.ab.ca).
Still, it is only by studying actual texts that we can learn about the use of metaphor. My data consist of 90 innovative metaphors, which are very unevenly distributed in the 11 text samples. In some of the sample texts there are only a few metaphors; by contrast Julian Barnes has 19, Margaret Drabble 14, and P. D. James 11 instances. This means that individual variation is quite considerable, disregarding ªction categories.
2.1 Some limitations of corpora Any corpus suŸers from some limitations. One is the use of text fragments. The text samples in the Oslo corpus make up chunks of some 15,000 words, taken from the beginning of the texts. The advantage of using such passages is that they are self-
248 Kay Wikberg
explanatory; i.e. the text world is built up from scratch. However, when studying metaphors one would ideally like to have complete texts since metaphors establish both propositional and attitudinal meaning; similarly, they can contribute to coherence and a narratological perspective (Werth 1999) if they form semantic and/ or associative networks which extend throughout a text. The obvious eŸect of the limited size of the texts in this study is that the metaphors can only be studied in their local functions, i.e. in practice within the sentence or the paragraph. Still, some metaphors relate to the theme or structure of a novel, as in (6) below, an example from Margaret Atwood’s Cat’s Eye. Owing to the limited size of the corpus samples, it is just by chance that certain metaphors are found, and there is no way of knowing what signiªcance a particular metaphor has in relation to the complete text. To be able to ªnd what I consider active (Goatly 1997) or innovative metaphors in the corpus I had to adopt a bottom-up approach, which involves reading the text. This is a prerequisite for any serious study of active metaphors, and also conventional metaphors since they can be revitalized if used by good writers, which means that their impact cannot always be understood fully unless seen in relation to other neighbouring ªgurative expressions. In poetic language we are used to ªnding clusters of metaphors representing semantic ªelds. Imaginative ªction, too, often contains systems or networks of metaphors, which therefore tend to be thematically relevant, or metaphors can appear in ‘bursts’ which stretch over a sequence of sentences. As regards translation corpora, there is an additional limitation: the dependence on the translator. This is more serious when dealing with unique metaphorical expressions than in research on most other linguistic features, which tend to occur with some frequency. In a previous study of mine it turned out that translators occasionally failed to translate the English metaphor into one or two of the languages (Wikberg 2003). In several cases there was no linguistic justiªcation for this. Although the translators were known, each text in the corpus had been translated by a separate translator, and therefore there was little point trying to look for idiosyncratic translation strategies. Translators work under pressure, which of course may be one explanation for their omitting textual material.
3.
On the translation of metaphor
In the study just referred to, using data from the same corpus, I show that the translation of innovative metaphors from English into Norwegian, Swedish and Finnish is mostly based on equivalent images and that most of the changes that occur in the rendering of the metaphors are simply due to the constraints of the target
English metaphors and their translation 249
languages. Problems like those illustrated in (1) were actually exceptional; most of the translations followed the originals fairly closely. As a non-Indo-European language it is obvious that Finnish has many structural characteristics which contrast with the other languages in the corpus. Thus similarity relations, which make up the main underlying semantics of metaphors, can sometimes be expressed by very diŸerent means, as in (2) her eyes seemed dark as dark (frozen simile) N øynene hennes virket mørkere enn mørkest (‘darker than darkest’) F … tummaakin [partitive + particle] tummemmat [comparative] (“even darker than dark”; JB1.3.s61) Important observations on the translation of metaphor have been made by scholars such as Dagut (1976), Newmark (1981, 1985), and Snell-Hornby (1988). Dagut focuses on the relation between language-speciªc properties, cultural in¶uence, and metaphorical expressions: the framework of ‘possible’ metaphors for any given language is determined by a combination of the accumulated cultural experience of the members of that language-community and the ‘institutionalized’ semantic associations of the items in their lexicon (either separately or in combinations). (p. 32)
Newmark (1985), who by comparison is more optimistic in his view of translatability, lists seven procedures for translating metaphors, sets up a useful typology of metaphors and examines the translation of each category with examples from French and German. Like Dagut, Snell-Hornby is concerned with the culturespeciªc aspect of translation. She thinks that translatability is dependent on both the structure of the metaphor and its function in a given text: Whether a metaphor is ‘translatable’ (i.e. whether a literal translation could recreate identical dimensions), how di¹cult it is to translate, how it can be translated and whether it should be translated at all cannot be decided by a set of abstract rules, but must depend on the structure and function of the particular metaphor within the text concerned. (Snell-Hornby, 1988: 59)
An example which illustrates what Snell-Hornby is referring to is (3) So Tad Trenton, four years old, lay in his bed, all wires and stiŸ Erector Set braces (SK1.1.s98) N Der lå ªre år gamle Tad Trenton i sengen, stiv som en pinne. (‘stiŸ as a pin’) S … spänd som en fjäder (‘spring’) och stel som en pinne. F … kauhusta [ELATIVE] kankeana [ESSIVE] (literally ‘for fear stiŸ’)
250 Kay Wikberg
An Erector set consists of construction toys made of metal, thus “stiŸ”. Understandably, all the translators dropped a reference to this object and only the Swedish version rendered the Vehicle all wires (spänd som en fjäder “wound up like a spring”). A few sentences later King refers to wires and Erector Set once more: (4) But little by little the wires unsnarled themselves and stiŸ Erector Set muscles relaxed. (SK1.1.s103) N Men litt etter litt løsnet spenningen i kroppen (‘relaxed the tension in the body’) S … började trådarna trassla upp sig; … F … jännitys kuitenkin hellitti (literally ‘tension still relaxed’) The Swedish version tries to be true to the source language in the ªrst clause, maintaining the concrete reference to metal, but the problem is that trådarna (“the threads”) does not convey the idea of a spring (fjäder) presented in (3) when combined with a verb like trassla upp sig. Threads can be unsnarled or disentangled but hardly wound up. The result is that the Norwegian and Finnish translations, which just refer to bodily relaxation, seem more successful. These instances, then, may serve to illustrate the importance of preserving the source language Vehicle in the target language when a metaphor ties in with and develops a previous metaphor. Another thing that these two examples illustrate is that dropping reference to a culture speciªc phenomenon need not involve much semantic loss. Like Erector, the trade name Playtex (a kind of rubber gloves) was also dropped in the translations. Metaphors can be semantically imprecise and are sometimes ambiguous by nature since they refer to relations. Such semantic indeterminacy may actually be an advantage for the translator, provided the target language metaphor preserves the semantic domains and is in keeping with the text world that has been built up to a given point in the discourse and the expectations created in the reader. Thus, even if there may be diŸerences in the way in which the target language structures the metaphorical elements, this need not necessarily spoil the ªgurative eŸect. Further examples of metaphors with comments on their translations will be given in the next section, in which we shall also be considering metaphor theory more generally.
4.
Three interpretations of ‘metaphor’
Let us begin by making a distinction between three uses of the term ‘metaphor’: – – –
metaphor as linguistic expressions (4.1); metaphor as a cognitive concept (4.2); metaphor as a discourse element (4.3).
English metaphors and their translation 251
The diŸerence between these is not always made clear in the literature. Thus a lot has been written about the ªrst two whereas less is still known about the communicative functions of metaphor in discourse.2 It is worth noting that many metaphorical expressions can be regarded as realisations of underlying conceptual metaphors. At the same time they occur in discourse, not just in isolated sentences. While linguistic analyses of metaphors tend to focus on expressions within the sentence, literary scholars have always studied metaphors in their context, for example how clusters of related images and metaphors contribute to thematic development. It is important that a linguistic approach to metaphor also pays due attention to the textual and communicative aspects of metaphor and aims at the inclusion of metaphor in an overall discourse model.
4.1 Metaphor as linguistic expressions Metaphors vary greatly in formal and semantic complexity. If we include conventional or lexicalized metaphors, formal variation covers anything from single words as in (5) Her Cape: the small houses of District Six all huddled together (ABR1.1.1.s451) N Hennes Cape: de små husene i Distrikt Seks som alle kryper sammen. S Hennes Kap: De små husen i Sjätte distriktet tätt hopkurade F Hänen Kapinsa: District Sixin pienet, yhteen pakkautuneet (‘packed together’) talot to sequences of expressions which extend over a number of sentences as in (6) Time is not a line but a dimension, like the dimensions of space.. . But I began then to think of time as having a shape, something you could see, like a series of liquid transparencies, one laid on top of another. You don’t look back along time but down through it, like water. Sometimes this comes to the surface, sometimes that, sometimes nothing. (MA1.1.1.s6) (5) represents a conventional, lexicalized metaphor with local impact whereas (6) bears on the structure of the whole novel, which describes the main character both as a girl and as a grown-up woman. Atwood shows us both by shifting the transparencies. The central problem that arises when studying metaphor is obviously how to identify the relevant expressions. What counts as a metaphor? The problem is in part much the same as when looking for units having speciªc communicative functions such as speech acts and moves in face-to-face conversation. Innovative metaphors are instances of language use and have to be interpreted in context.
252 Kay Wikberg
Similarly, the interpretation of a speech act is possible only in its textual context. However, there is no single formal feature that can help us to pin down a metaphor. The crucial factor is rather semantic, the presence of a nonliteral expression. Since there are other kinds of nonliteral expressions as well such as idioms, irony, understatement, hyperbole, and so forth, this is not enough. A very general deªnition of metaphor which could serve as a starting point would be this: Metaphor is a similarity relation between two semantic domains. (Miller, 1993) This deªnition, which presupposes propositional analysis, focuses on meaning which underlies the surface structure.3 Following Miller, Steen (1999: 88) argues that the starting point for linguistic analysis should be “the conceptual analysis of propositions”. For the comparison of translations this obviously raises the interesting question of diŸerent levels of abstraction in the analysis. Another question is whether it is always possible to envisage the existence of underlying propositions for all sorts of metaphors. In this paper I shall ignore the propositional level. As regards the semantic domains referred to in Miller’s deªnition, they must somehow be comparable so that the reader or hearer can make use of processing procedures like implicatures and inferencing to bridge the gap between what is said (written) and what is implicated. The standard linguistic explanation has been to see the relation that underlies a metaphor as a transfer of features or attributes from the Vehicle to the Topic. Cognitive linguists speak of a mapping of a Source domain onto a Target domain. A simple example is (5), in which huddle colligates with an [INANIMATE] Subject, contrasting with a conventional [ANIMATE] one. This animistic eŸect on the Topic is lost in the Finnish translation, which also reduces the metaphorical impact. Instead, the Finnish version conveys the idea that the houses had been packed together in a small space. Whereas the semantic feature approach applies to relatively simple metaphors, the range of metaphorical expressions is too wide always to allow an analysis in such terms. For example, both analogies and as if-similes involve more elaborate semantic relations, as illustrated in (7): (7) Stuart on a double-date is deªantly cognate with a breadstick still in its wrapper. (JB1.2.3.s59) N Stuart som dobbeltstevnemøtedeltager er uavvendelig beslektet med et kuvertbrød som fremdeles ligger i forpakningen sin S Stuart på en dubbelträŸ är konstitutionellt besläktad med en tandpetare (‘toothpick’) i oöppnat fodral. F Stuart on tuplatreŸeillä yhtä (‘as’) järisyttävän [genitive] aktiivinen [ground] kuin (‘as’) suolatikku sellofaanikääressä.
English metaphors and their translation 253
The analogy in the original seems somewhat far-fetched although the reader certainly infers that Stuart is inexperienced and that he therefore is unlikely to cope with a double-date. The source language expression be cognate with is simply an unusual way of expressing resemblance. The Finnish translator turns this into a simile and bases his translation on irony and the imagined (in)activity of a breadstick in its wrapper as being a parallel to Stuart’s expected behaviour. From the point of view of translation, the Finnish version is interesting in the sense that it adds a Ground element, which has to be inferred in the other target languages. Finally, the Swedish translator compares Stuart to a ‘toothpick’, which is deªnitely wrong.4 Recent work by Cameron (1999) gives us a more solid basis for the interpretation of metaphor than just relying on semantic features, which may be hard to pin down anyway. According to Cameron, for a metaphor to be an active (creative) metaphor it would have to be characterized by a high degree of the following factors: – – –
–
incongruity between Topic and Vehicle; vitality, i.e. form a novel link between Topic and Vehicle; cognitive demand: a more complex dimension having to do with both Topic and Vehicle, their abstraction, generality and complexity connotative power in the Vehicle.
A good example which satisªes these criteria is (1). Incongruity is a sort of lack of ªt or semantic mismatch. There must be a certain distance or incongruity between the two domains, at the same time as they must be close enough to allow analogical reasoning. This certainly applies to (7), for instance. The second notion, ‘vitality’, is related to frequency, the fact that the Topic-Vehicle linkage is an unusual combination. Many metaphors can be looked at as unusual colligations or collocations, carrying nonliteral meaning. ‘Cognitive demand’ ªnds support in Relevance Theory (Sperber & Wilson 1986), according to which complex and abstract metaphors may require additional cognitive eŸort to be understood. Finally, ‘connotative power’ is connected with what Dagut refers to as ‘culturally shared associations’ elicited by an item or an image, as in (12). When it comes to the use of corpus methodology for the study of metaphors, it is obvious that it is the surface elements that count. Only lexicalized metaphors and metaphors that are clearly signalled by special markers can be spotted without manipulating the text. Figurative uses of polysemous words make up one category (e.g., the brow of the hill, the mouth of the closet). Provided the item is known, such uses of conventional metaphors can easily be detected by means of concordancing programmes and have already been registered, for example in the Cobuild study by Deignan (1995). They are found in dictionaries, including bilingual dictionaries,
254 Kay Wikberg
which can tell us whether e.g. diŸerent languages use names for body parts or colour diŸerently for ªgurative purposes. According to Simon-Vandenbergen (1993), it is important in metaphor research to include such conventional lexicalized metaphors and not just creative ones. This is because conventional metaphors can regain vitality through their recurrence and their link with others from the same domain. On the other hand, they form the background against which creative metaphors are formed and are to be interpreted. (Simon-Vandenbergen 1993: 180)
Cacciari (1998) adopts a similar position. The main reason why this study focuses mainly on creative metaphors is that they are cognitively and stylistically more interesting, but an analysis of how ªgurative language is used in a short story or novel should not ignore lexicalized metaphors. However, Stern (2000: 28) makes the important point that “the liveliness (in one sense) of a metaphor is at least in part a function of its degree of dependence on its context”. He also argues convincingly that it is not the expression as such but its interpretation in a given context that is decisive for the distinction between living or dead metaphors. As to the structure of the less conventional metaphorical expressions in my corpus, clauses with ªnite verbs, predicate metaphors, make up the most common category of Vehicles, some 70 per cent of all instances. Typically, quite a few verbs convey speciªc manner components (cf. loom, fester, supporate, spill from, huddle together, glower away, chug away, excavate, liquidize, burrow into, bottle up). Most of these verbs are intransitive, but what is more important is that they have a descriptive function. This is signiªcant when considering the communicative impact of metaphor. Katz (1998: 22) refers to a paper by Torreano and Glucksberg according to which, “the metaphoricity of a sentence is directly related to the degree to which the verb is used in thematically unusual ways”. Liquidise, as in she can liquidise you like a carrot in a kitchen blender (RD1.7.s34), is certainly a transitive material process verb, enough to remind us of what happens if somebody is put through a food processor. The like-phrase adds precision and the whole clause can be classiªed as a hyperbole. Several nominal metaphors which generate vivid images have already been discussed. According to Goatly (1997: 83), nouns refer to things and phenomena which are more easily perceptible than verbal processes and “yield richer interpretations than V-terms [= Vehicle terms] of other word classes”. As he also points out, the A is B formula, which is so often used to represent typical metaphors, has to be ªlled by Noun Phrases: (8) The pound is a goldªsh swimming in a tank of piranhas (ST1.1.9.s179) N pundet svømmer som en gullªsk i et basseng med pirajaer (‘the pound swims like a goldªsh…’)
English metaphors and their translation 255
S Pundet är en guldªsk som simmar i en tank med pirayor F Punta on kultakala, joka ui piraija-akvaariossa [compound] For the translator (8) is a simple construction. However, as the example shows, the Norwegian version lacks the equative be (N være) and thus also the subclause. Instead the Ground swim is upgraded to a ªnite verb followed by an adverbial simile, which obviously would have been an equally good alternative in the other target languages as well. The eŸect of choosing the Norwegian solution is to focus slightly more on the behaviour of the pound, at the same time avoiding the contradiction conveyed by the other three versions. Similes are often mentioned as a separate category of ªgurative expressions and I shall therefore discuss them separately below. Apart from similes, other speciªc markers are rare in the present corpus. However, some instances are found in Margaret Drabble’s The Middle Ground: (9) She was fond of speaking to the girls, especially as they grew older, of the perils of sex, and unfortunately this theme seemed to draw her unconsciously into the strangest realm of sexual imagery — she would speak of guns and triggers, of climaxes, of ªlling the girls in and of playing hall. (MD1T.1.s402) N … trekke henne helt ubevisst inn i en forunderlig verden av seksuelle fantasier; hun snakket gjerne om pistoler og avtrekkere, om klimaks, om å fylle ut pikene og om å spille ball. S föra henne in på olyckliga metaforer av alla de slag — hon talade om pistoler och om avskjutning, om orgasmer, om att hålla stånd och om att pungslå. F … hän puhui pistooleista ja liipaisimista, huipennuksista, täyttymyksistä ja pallonpeluusta [abbreviated] (10) So by a general consensus the party, as it were, metaphorically turned its back on her (MD1.1.s698) N … metaforisk … [as it were Ø] S … bildligt talat …[as it were Ø] F … ikään kuin metaforisesti … (11) And it was at this point that Kate rose, as it were, from the crumbs, and did her stuŸ. (MD1.1.s706) In (9) we are dealing with metalanguage within the novel, signalled by the general noun imagery. The other examples contain more ordinary markers.
256 Kay Wikberg
Similes In this paper no rigid distinction is made between metaphors and similes. The main diŸerences between a metaphor and a simile are that a simile always has a marker (see below) and that it “makes a weaker claim” (LakoŸ & Turner 1989: 133) than a metaphor. Thus a simile can be metaphorical on the condition that the TopicVehicle combination is somehow novel and adds relevant information. We have already seen several examples ((1), (3), (7)) in which a source language metaphor has been translated as a simile, which is what one might expect, considering the simile’s more explicit nature. About 25 per cent of the examples of creative metaphors in this corpus contain either a phrasal or clausal simile, as in the following two instances. (12) is a description of the Whistler’s methods of killing his victims, in P. D. James’s detective story Devices and Desires: (12) [And then he cut oŸ their hair and stuŸed it in their mouths,] like straw out of a Guy on 5 November (PDJ3.1.1.s90) N … lik halmen som stakk ut av en Guy Fawkes-dukke 5. november. S … så att det stack ut som halmen ur en Guy-Fawkes-docka den 5 november F … niin että he olivat kuin olkia pursuvat Fawkesnuket 5. marraskuuta. Here the phrasal simile constitutes the metaphor. Not surprisingly, all three translations in (12) add an explanatory word for Guy (N dukke, S docka, F nukke = ‘doll’), which can be seen as a type of necessary ‘explicitation’. The simile can be regarded as metaphorical because of its cultural connotations and the link hair (Topic) – straw (Vehicle); it is certainly more than just a Noun Phrase (cf. your friend talks like a dictionary or the dog was going to bite oŸ Ted’s head like a lollipop) and the source language sentence would be incomplete if it stopped at straw. The simile expresses the result of the killer’s act and in all the target languages there is a verb which refers to the achieved eŸect (N stakk ut, S stack ut ur, F pursuvat). In the Swedish version the result is made explicit by the resultative så att. Finally, (13) contains the lexicalized metaphor ¶ooded, but the clausal as if-simile5 enhances its metaphoricity since it adds an appropriate analogy (money – ¶ooded : dam – burst): (13) Money had ¶ooded out of the country, as though a dam had burst (ST1.1.5.s47) N Pengene hadde strømmet ut av landet som om en demning var blitt sprengt. S Penger hade strömmat ut ur landet som en fördämning hade brustit F Rahaa oli virrannut maasta pois kuin pato olisi murtunut
English metaphors and their translation 257
Moreover, (13) illustrates the conceptual metaphor money is liquid. The additional simile has the eŸect of exemplifying how the ¶ooding could have taken place.
4.2 Metaphor as a cognitive concept LakoŸ (1993) argues that metaphor is not a property of language at all; it is rather a property of our conceptual system. But when we look at individual languages, we shall ªnd that there are diŸerences in the ways in which the cognitive models surface in terms of lexico-grammatical expressions. As we have seen already, for the present study the only possible method was to start out from actual expressions rather than abstract cognitive models. Still, linguistic evidence for quite a few conceptual metaphors was easy to ªnd in the data and in all the languages studied such as words = liquid, emotion = liquid, life = a gamE, and body = building. If we exclude the treatment of metaphor as a deviant linguistic expression in early transformational grammar, and as a cognitive phenomenon in cognitive linguistics, metaphor has not attracted much attention among theoretical linguists. According to Leech (1974), metaphor can in part be explained in terms of transfer features. This view of metaphor as a transfer phenomenon is still valid and has already been illustrated in (5). Another example is (14), which shows the shift from [abstract] to [concrete], one of the most common metaphorical processes: (14) [And what more practical way of repaying them than to ensure that] their own little seedling of bonheur had time to sprout and shoot, to root and burgeon? (JB 1.3.3.s29) N sørge for at deres egen lille kim av bonheur ªkk tid til å spire og sette skudd, slå rot og blomstre S se till att deras spirande bonheur ªck tid att slå rot och frodas, skjuta skott och knoppas? F varmistaa, että heidän hennon bonheurinsa [-nsa = poss suffix] taimella oli aikaa kasvaa ja versoa, juurtua ja kehittyä. This example also demonstrates the interesting fact that the French word bonheur has been transferred to all three languages, even to Finnish with a possessive su¹x. With reference to (14) and the theory of conceptual metaphor one might say that we are dealing not only with a concretizing metaphor but also with a root analogy: state of mind = young plant. Conceptual metaphor is a powerful theory and is undoubtedly here to stay. It provides an experiential basis for explaining the existence of metaphorical expressions which fall into speciªc patterns. Without the cognitive-linguistic theory of metaphor linguistic analysis of metaphor tends to become purely descriptive.
258 Kay Wikberg
LakoŸ (1993: 246) lists various kinds of evidence in favour of this approach, such as polysemy, typical ways of inferencing, generalizations about semantic change, and psychological experiments. We must, however, allow for possible cultural contrasts in the way conceptual metaphors are formed and expressed since norms and beliefs can vary from one culture to another (Quinn 1991). A number of theoretical and methodological issues have also been raised about conceptual metaphor theory, most recently by McGlone (2001).
4.3 Metaphor as a discourse phenomenon Metaphor is a pragmatic phenomenon, but at the same time its cognitive and semantic character shows up in the way lexico-grammatical expressions are combined, often in new colligations or collocations, to convey meanings which may lie beyond the capacity of literal language. The question arises how to handle metaphor and ªgurative language in an overall grammatical model. In Principles of Pragmatics, Leech discusses the relation between formalism and functionalism and looks at pragmatics in relation to Halliday’s three well known metafunctions: ideational, interpersonal, and textual. One of Leech’s postulates is that, “Grammar is ideational; pragmatics is interpersonal and textual.” (1983: 56) There is certainly support for the view that metaphors are interpersonal and/or textual if one examines how metaphors are used in imaginative prose. At the same time I would like to argue that Halliday’s concept of ‘grammatical metaphor’ (1994: 340–367) is not adequate for dealing with the full range of metaphorical expressions. For example, in many instances it is hard to think of an unmarked or ‘congruent’ realization of a complex metaphor. However, the major functions of metaphors can still be seen to ªt into what in systemic-functional grammar is referred to as interpersonal and textual metafunctions. Werth (1999: 190–194) distinguishes between diŸerent kinds of ‘functionadvancing propositions’. Thus reporting the sequence of actions or events in a narrative would be ‘plot-advancing’. New referents are introduced and things happen to the characters as time rolls on in the story. Very few of the metaphors in the data are used for such purposes. If they are, they tend to re¶ect the narrator’s attitude and represent lexicalized metaphors rather than innovative ones. Since novels deal with characters, metaphors and other nonliteral expressions are well suited to depict their mental states in more or less subjective ways. The overwhelming majority of the innovative metaphors in the corpus describe rather than narrate. As we have already seen, there is usually less scope for subjectivity in detective stories, in which the reader expects the plot to be unravelled rather than being given a subtext which is the narrator’s or author’s way of going beyond facts into imaginative worlds.
English metaphors and their translation 259
Functional grammar is not incompatible with Paul Werth’s theory of textworlds, which adds a necessary cognitive dimension and a means of helping us to understand how metaphor is interpreted as part of the text: What it [metaphor] does is to provide a sub-text which sheds light on the topic of discourse, and the one who provides this sub-text is usually the author, and not a character. .. we can see that it works by opening up an area of experience in terms of which the discourse topic can be (partially) interpreted. (Werth 1999: 323–324)
Werth calls this ‘double vision’. To take an example which lies outside the present corpus because it requires a complete text, in Margaret Atwood’s Surfacing there are metaphors which repeatedly contrast the past with the present, as when the narrator describes her parents’ conservatism and remoteness as illustrated in (15) and (16): (15) I thought of them as living in some other time, going about their own concerns closed safe behind a wall as translucent as jello, mammoths frozen in a glacier. (ch. i, p. 6) (16) They were from another age, prehistoric, when everyone got married and had a family, children growing in the yard like sun¶owers; remote as Eskimos or mastodons. (ch. xvii, p. 209) An interesting fact about these metaphors is that they cluster in sentence ªnal position where they concretize and develop an earlier text element: in some other time > mammoths; prehistoric > mastodons At an interdisciplinary workshop on corpus-based and processing approaches to ªgurative language at Lancaster University in 2001 it was suggested that it might be possible to spot some occurrences of metaphorical expressions in corpora by examining unique words in word lists. Mammoth and mastodon are such words, winterwhistle (as in (1)) another. Goatly (1997: 164) argues that “nouns which refer to imaginable things have a particular vitality” and tend to generate more images than other word classes. The nouns mentioned certainly belong to this category of words. Such unique items contribute to cohesion only when combined with other more high-frequency items belonging to the same semantic ªeld.
4.3.1 On the pragmatic functions of metaphor The orthodox view is that metaphors are descriptive, in literature often ‘decorative’. That of course is a crude simpliªcation because languages also have other means of describing. If a metaphor is used instead of a literal expression, the author must be describing with a purpose and has made a deliberate choice. Goatly (op.cit.) discusses a number of functions that metaphors might have, including ‘ªlling lexical
260 Kay Wikberg
gaps’, ‘explanation’, ‘expressing emotional attitude’, ‘decoration, disguise and hyperbole’, ‘textual structuring’, and ‘foregrounding’. In the Oslo corpus it is easy to ªnd metaphors which express emotional states or attitudes, and metaphors which contribute to vivid description or textual structuring. Actually, we are often dealing with more than one function simultaneously, as in this example from André Brink’s The Wall of the Plague: (17) The stern black slate roofs of Auvergne, houses like nuns with their backs turned to the light. (ABR1.1.1.s316) N De strenge, svarte skifertakene i Auvergne, hus lik nonner med ryggen mot lyset S Auvergnes stränga, svarta skiŸertak, hus som nunnor med ryggen vänd mot ljuset F Auvergnen karun mustat liuskekatot, talot kuin selin valoon kääntyneitä nunnia The simile gives a more vivid description than the preceding Noun Phrase (the stern … roofs). Like (5), which also describes houses, this metaphor is clearly emotive. Some metaphors in ªction reappear elsewhere in the text although with a slightly diŸerent make-up: either the Topic or the Vehicle may have changed but in the company of items from the same lexical set. This is where the computer might help to spot the positioning of reoccurring elements throughout the text. Provided the text contains references that allow page numbering or even more precise position of the construction, it may be possible to ªnd clusters of semantically rich word combinations and their locations. A multilingual corpus will in this case allow the researcher to compare partial semantic ªelds contrastively. Metaphors in restricted domains and genres make up a ªeld in which the computer can be used even more easily, simply because there is a much narrower range of expressions (cf. Andersen 2000). We are just about beginning to learn about how metaphors are used in diŸerent genres. But even within the same genre there can be great diŸerences in the use of metaphorical expressions. DiŸerent authors may use metaphors diŸerently and metaphors may be diŸerently distributed throughout a text.
5.
Idioms
The relation between metaphor and idiom has attracted a great deal of attention recently (e.g. Gibbs 1994; Moon 1998; Glucksberg 2001; Wikberg 2001). It has long been recognized that many idioms convey metaphorical meaning and that they are based on vivid expressions which have been lexicalized. Gibbs presents a conceptual view of idioms contrasting with the traditional view of idioms as ‘dead’
English metaphors and their translation 261
lexicalized entities. He argues that there are many idioms that are motivated by ªgurative thinking and that familiar idioms are accessed as units without being decomposed. Moon classiªes noncompositional ªxed expressions into metaphors according to degrees of transparency. She also provides valuable information on the functions of metaphors, such as the proportion of informational versus evaluative and positive versus negative evaluation. Glucksberg looks particularly at two properties of idioms, compositionality and formal variability, but he is also interested in the relation between that variability and the pragmatic and communicative functions of idioms. Finally, Wikberg compares metaphors and idioms in terms of eight diŸerent dimensions, including ªxedness, paraphrasability, translatability and communicative functions. What, then, are the consequences for idioms in translation? Many problems with idioms in translation are well recognized, the most obvious one being references to culture-speciªc phenomena, as in (18b) below. Opaque idioms do not translate easily since they are di¹cult to decode. Even so they may correspond to semantically equivalent expressions in the target language and result in translations having much the same eŸect as the original. In a study of four English novels and their translations into Danish, Gottlieb (1995) found that about 80 per cent of the source language idioms were rendered by semantically equivalent expressions but only 26.5 per cent were identical idiomatic expressions. Since the idioms occurring in the present corpus seemed to have been more di¹cult to translate than innovative metaphors, I decided to examine how a sample of idioms had been translated. It turned out that of a set of 18 clear-cut idioms only three had been translated by means of idioms in all three languages: (18) a. pin sb. in a corner b. not let X near with a ten-foot pole c. go to pieces These three expressions all contain concrete references in their literal senses and their idiomatic senses are relatively transparent. Thus they diŸer from such idioms as stand the cheek, be at odds with, cut zero ice with, and be a smartmouth. This might explain why they caused less di¹culty than these more abstract idioms. The fact that (18b) did not have a direct equivalent in the target languages is not very surprising. A ten-foot pole is not a concept that would be lexicalized in Nordic languages, and consequently the tool was rendered with reference to other objects (N ildrake = ‘poker’; S tång = ‘tongs’, F tikku = ‘pin’). Additional corpus data show that (18b) is more ¶exible than dictionaries suggest. The form given in Cambridge International Dictionary of Idioms is I wouldn’t touch sb/sth with a ten-foot pole. The British National Corpus has one instance of this basic form but also I wouldn’t rape you with a ten-foot pole.
262 Kay Wikberg
6.
Conclusion and perspectives
For a study of metaphors and idioms in translation to be meaningful and more than just a list of annotated examples we need a theory which comprises the linguistic, conceptual and communicative aspects of such nonliteral expressions. Elements of a comprehensive theory are slowly emerging. Metaphor research does not consist merely of ‘deconstruction’ of metaphorical relations in terms of the established elements Topic, Vehicle, and Ground and their explicit or even implicit exponents. Admittedly, interaction theory, comparison theory (Goatly 1997) and class-inclusion theory (Glucksberg 2001) deal with the relation between Vehicle and Topic, but they do not cope with relations beyond the sentence or relating metaphors to text worlds. This is where functional approaches may be useful. Most of the metaphors in the present corpus are evaluative in some respect and would therefore, at clause level, realize Halliday’s interpersonal metafunction. For the analysis of metaphor beyond the clause or sentence, Steen’s communicative analysis of metaphor (1999a, 1999b) contains elements which could be handled by Halliday’s textual metafunction, notably lexical cohesion, but most of them are best catered for by discourse analysis and literary theory. What speciªcally are the implications for the study of metaphor in translation? The crucial thing for correct rendering into the target language is obviously understanding of the original metaphorical expression and its pragmatic function. In spite of the subjective character of literary discourse the current corpus, data show remarkable equivalence between source language metaphors and target language translations. The problems that arise for cultural reasons are in this case relatively uninteresting. More interesting are the semantic ªelds that come into play in the metaphorical expressions and their interpretation. Ideally the translator (and researcher) should have access to a word list for the original, and a thesaurus and a collocation dictionary for each language. Such tools would make it easier to explore possible diŸerences in the semantic ªelds and the collocational proªles of words. The great advantage of having access to a multilingual corpus like the Oslo corpus is that it provides authentic data, original and translated. The availability of the same texts in a number of diŸerent languages is an asset for the linguistic research community. Contrastive diŸerences work bottom-up, from the lexical level to the sentence level and beyond. Once a lexical choice has been made, it may have repercussions elsewhere in the text. The computer helps to trace chains of cognitively related metaphors. Access to a large control corpus like the British National Corpus can be an advantage since it enables researchers to ªnd speciªc instances of ªguratively used words and idioms and also elements of innovative metaphors belonging to the same genre with information on the most common collocates. Finally, it is obvious that, for the study of genre diŸerences such as
English metaphors and their translation 263
between general ªction and detective stories, the present subcorpus is too small. The diŸerences that emerge in the use of metaphors are rather dependent on the authors of the respective texts.
Notes * An earlier version of this paper was presented at The Corpus Linguistics Conference at Lancaster University, March 30–April 2, 2001. 1. Websites: http://www.hf.uio.no/iba/prosjekt/ and http://www.hf.uio.no/german/sprik/ english/ ) 2. An exception is Steen (1994, 1999a, 1999b). 3. Searle (1981: 269-272) criticizes a previous version of Miller’s similarity thesis. 4. Julian Barnes, the author of the extract, accounts for a higher number of metaphors than any other author in the extracts analysed. Elsewhere Barnes also seems to be indulging in using rare words. 5. For a description of as if (though)-clauses, see Wikberg (1999).
References Andersen, Mette Skovgaard. 2000 “Metaphor matters”, Hermes, Journal of Linguistics 24: 57–80. Atwood, Margaret 1992 (1973) Surfacing. London: Bloomsbury Classics. Cacciari, Cristina 1998 “Why do we speak metaphorically? Re¶ections on the functions of metaphor in discourse and reasoning”. In A. Katz et al., 119–157. Cambridge International Dictionary of Idioms 1998 Cambridge: Cambridge University Press. Cameron, Lynne 1999 “Identifying and describing metaphor in spoken discourse data”. In Researching and Applying Metaphor, L. Cameron and G. Low (eds), 105–132. Cambridge: Cambridge University Press. Cameron, Lynne & Graham Low (eds) 1999 Researching and Applying Metaphor. Cambridge: Cambridge University Press. Dagut, M. B. 1976 “Can ‘metaphor’ be translated?”. Babel 22 (1): 21–33. Deignan, Alice 1995 Collins Cobuild English Guides 7: Metaphor. London: HarperCollins. 1999 “Corpus-based research into metaphor”. In L. Cameron and G. Low (eds), 177–199.
264 Kay Wikberg
Gibbs, Raymond W. 1994 The Poetics of Mind. Figurative Thought, Language and Understanding. Cambridge: Cambridge University Press. Glucksberg, Sam 2001 Understanding Figurative Language. From Metaphors to Idioms. Oxford: Oxford University Press. Goatly, Andrew 1997 The Language of Metaphors. London and New York: Routledge. Gottlieb, Henrik 1995 “Oversætteres fornemmelse for idiomer. Et punktnedslag i oversættelsen af litterære tekster”. In Oversættelse av litteratur [ Danske Afhandlinger om Oversættelse nr. 6], P. Florentsen (ed.), 105–125. Copenhagen: Copenhagen University. Halliday, Michael 1994 Introduction to Functional Grammar. 2nd ed. London: Edward Arnold. Katz, Albert N. 1998 “Figurative language and ªgurative thought: a review”. In A. Katz et al., 3–43. Katz, Albert. N., Cacciari, Cristina, Gibbs, Raymond W. and Turner, Mark 1998 Figurative Language and Thought. New York & Oxford: Oxford University Press. LakoŸ, George 1993 “The contemporary theory of metaphor”. In Metaphor and Thought. 2nd ed., A. Ortony (ed.), 202–251, Cambridge: Cambridge University Press. LakoŸ, George and Turner, Mark 1989 More Than Cool Reason. A Field Guide to Poetic Metaphor. Chicago: University of Chicago Press. Leech, GeoŸrey 1974 Semantics. Harmondsworth: Penguin. 1983 Principles of Pragmatics. London: Longman. Miller, George A. 1993 Images and models, similes and metaphors. In Andrew Ortony (ed.), Metaphor amd Thought. 2nd ed. 357–400. Cambridge: Cambridge University Press. McGlone, Matthew S. 2001 “Concepts as metaphors”. In S. Glucksberg, 90–107. Moon, Rosamund 1998 Fixed Expressions and Idioms in English. A Corpus-Based Approach. Oxford: Clarendon Press. Newmark, Peter 1981 Approaches to Translation. Oxford etc.: Pergamon Press. 1985 “The translation of metaphor”. In The Ubiquity of Metaphor. Metaphor in Language and Thought, W. Paprotté and R. Dirven (eds), 295–326. Amsterdam/Philadelphia: John Benjamins.
English metaphors and their translation 265
Quinn, Naomi 1991 “The cultural basis of metaphor”. In Beyond Metaphor. The Theory of Tropes in Anthropology, J. W. Fernandez (ed.), 56–93. Stanford: Stanford University Press. Sage, V. 1994 “Metaphor in literature”. In The Encyclopedia of Language and Linguistics. Vol. 5. R. E. Asher (ed.), 2456–2461. Oxford etc.: Pergamon Press. Searle, John R. 1981 “Metaphor”. In Philosophical Perspectives on Metaphor. M. Johnson (ed.), 248–285, Minneapolis: University of Minnesota Press. Simon-Vandenbergen, Anne-Marie 1993 “Speech, music and dehumanisation in George Orwell’s Nineteen Eighty-Four: a linguistic study of metaphors”. Language and Literature, 2 (3): 157–182. Snell-Hornby, Mary 1988 Translation Studies: An Integrated Approach. Amsterdam: John Benjamins. Sperber, Dan and Wilson, Deirdre 1986 Relevance. Communication and Cognition. Oxford: Basil Blackwell. Steen, Gerard 1994 Understanding Metaphor in Literature. Harlow: Longman. 1999a “Metaphor and discourse. Towards a linguistic checklist for metaphor analysis”. In L. Cameron and G. Low (eds), 81–104. 1999b “Analyzing metaphor in literature: With examples from William Wordsworth’s ‘I Wandered as a Cloud’”. Poetics Today 20 (3): 499–522. Stern, Josef 2000 Metaphor in Context. Cambridge, Mass. & London, England: The MIT Press. Torreano, L. and Glucksberg, S. 1996 “When are verbs metaphorical? The role of selection restriction violation”. Paper presented at 37th Psychonomic Society Meeting, Chicago, November. Quoted in Katz, A. N. Figurative language and ªgurative thought: A review. In A. Katz et al. 1998, 3–43. Werth, Paul 1999 Text Worlds: Representing Conceptual Space in Discourse. London: Longman. Wikberg, Kay 1999 “The style marker as if (though): A corpus study”. In Out of Corpora. Studies in Honour of Stig Johansson. H. Hasselgård and S. Oksefjell (eds), 93–105. Amsterdam – Atlanta, GA: Rodopi. 2001 “Alive and kicking – Metaphors versus idioms: a comparison”. In Language, Learning, Literature. Studies Presented to Håkan Ringbom. M. Gill, A. W. Johnson, L. M. Koski, R. B. Sell and B. Wårvik (eds), 183–198. Åbo Akademi. 2003 “Studying metaphors using a multilingual corpus”. In Meanings in Contrast: The Cambridge Papers. Vol. 2. K. M. Jaszczolt and K. Turner (eds), 109–123. Amsterdam/Philadelphia: John Benjamins. Wright, W. Huntington 1927 The Great Detective Stories. http://gaslight.mtroyal.ab.ca.
Index of names
A Akman 202, 214 Altenberg 82, 85 Andersen 260, 263 Anderson 90, 108 Anscombre 141, 154 B Bäcklund 5, 13 Baicchi 6, 8, 17, 26, 29, 33, 35, 36, 40, 43, 62 Bamford 5, 9, 113, 116, 136, 154, 200, 214, 215 Barth 22, 36 Barton 142, 154 Bathia 202, 204, 214 Beaugrande 2, 11, 25, 36, 39, 62, 107, 108 Berkenkotter 139, 154, 203, 211, 214 Biber 5, 11, 65, 71, 74, 85, 108, 178, 183, 226, 227, 233, 240 Blakemore 140, 154 Bloomªeld 3, 11 Bondi 4, 7–9, 136, 139, 140, 142, 149, 154, 170, 199, 214, 215 Breivega 199, 214 Brown, G. 1, 2, 3, 47, 62, 89, 90, 109, 162, 171 Brown, P. 158, 160, 170, 175 Bruti 6, 8, 30, 36, 39, 40, 48, 54, 62 Bublitz 90, 109 Button 115, 137 C Cacciari 254, 263, 264 Cameron 253, 263, 265 Carter 91, 109, 110
Chafe 213, 214, 223, 227, 229, 240 Channell 129, 136, 137, 174, 175, 178, 183, 196 Cherry 213, 214 Chomsky 1 Clark 89, 90, 109, 110 Cloran 68, 85 Coates 90, 109 Collins 221, 225, 227, 229, 239, 240 Connor 10, 11 Coulthard 3, 4, 11, 13, 137, 142, 155, 156, 214, 216 Crismore 199, 214 Crompton 175, 196 D Danes 79, 85 Dascal 153, 154 Davison 39, 62 DeCarrico 199, 214 Deemter 43, 62 Deignan 253, 263 Del Lungo 199, 214 Delªtto 43, 62 Di Fazio 23, 24, 36 Dickinson 90, 91, 109 Donegan 28, 36 Dos Santos 153, 155 Dressler 22, 25, 28, 36, 39, 56, 62 DuBartell 97, 109 Duchet 20, 37 Ducrot 141, 154 E Eckkramer 22, 36 Ehlich 43, 44, 52, 62
268 Index of names
Emonds 219, 240 Enkvist 65, 85, 110 Erdmann 221, 227, 229, 239, 241 F Fairclough 3, 12, 202, 214 Farnsworth 199, 214 Fetzer 202, 214 Fillmore 119–121, 123, 128, 136, 137, 155 Firbas 229, 241 Firth 2, 12 Flowerdew 199, 215, 216 Fodor 32, 37 Fox Tree 90, 109 Fraser 140, 141, 153, 155 Fries 68, 85, 86 G Gee 202, 215 Geluykens 229, 241 Genette 21, 22, 37 Gernsbacher 41, 46, 48, 50, 62, 90, 108–110 Givón 25, 28, 37, 44, 52, 55, 62, 86, 90, 91, 108–110, 220, 229, 241 Glover 121, 125, 137 Glucksberg 254, 260–262, 264, 265 Goatly 248, 254, 259, 262, 264 GoŸman 157–160, 164, 171 Gonzales 138 Goodwin 114, 125, 132, 137 Gottlieb 261, 264 Green 39, 62 Grivel 21, 37 Gundel 48, 50, 61–63 Gutwinski 43, 63 H Haiman 28, 37 Halliday 2, 3, 7, 10, 12, 25, 35, 37, 40–43, 47, 63, 65–69, 72, 73, 75, 76, 84–86, 91, 93, 95, 96, 107, 109, 124, 137, 148, 153, 155, 164, 171, 175, 176, 196, 202, 215, 221, 241, 258, 262, 264 Hanks 121, 123–125, 137
Hansen 199, 215 Harris 3 Hasan 2, 7, 12, 25, 35, 37, 40–43, 47, 63, 65, 75, 76, 85, 86, 93, 95, 96, 107, 109, 124, 137, 153, 155, 164, 171 Hasselgård 10, 65, 75, 80, 81, 85, 86, 265 Heath 125–127, 135, 137 Hedberg 48, 50, 63 Hindmarsh 125–127, 135, 137 Hoek 20–22, 37 Hoey 4, 8, 12, 91–93, 99, 107, 109, 140, 155, 156 Horn 238, 241 Hübler 173, 196 Huckin 139, 154, 203, 211, 214 Huddleston 219, 220, 222, 227, 229, 241 Hyland 113, 137, 139, 140, 149, 153, 155, 173, 175, 176, 195, 196, 199, 202, 203, 215 I Inchaurralde 35, 37 Ivanic 199, 202, 203, 213, 215 J Jacoby 138 Jakobson 32, 37 JeŸerson 4, 12, 242 Jescheniak 41, 46, 48, 50, 62 John 199, 216 Jucker 140, 155, 156 K Kaltenböck 7, 219, 239–241 Kaplan 10–12 Katriel 153, 154 Katz 32, 37, 254, 263–265 Kesik 40–42, 44, 47, 50, 61, 63 Koj 28, 37 Kress 200, 215 Kripke 32, 37 L LakoŸ, G. 173, 174, 189, 196, 256–258, 264
Index of names 269
LakoŸ, R. 153, 155, 160, 164, 171, 196 Langacker 35, 38, 227, 241 Lee 115, 137 Lenk 90, 109 Levinson 122, 125, 130, 136, 137, 158, 160, 170, 175 Lindemann 188, 189, 195, 196 Linell 90, 91, 97, 107, 110 Longacre 77, 79, 86 Lucy 154–156 Lutz 232, 238, 241 Lyons 12, 21, 38, 119, 120, 125, 137 M Mackenzie 229, 238, 242 Maillard 41, 63 Mair 239, 241, 242 Malamud-Makowski 141, 153, 155 Mann 7, 12, 110, 242 Marello 21, 38 Markkanen 173, 175, 196, 197, 213, 215 Martin 7, 11, 12, 52, 73, 91, 93, 96, 110, 215 Matras 119, 137 Matthiessen 67, 68, 75, 76, 84, 86 Mauranen 6, 173, 177, 183, 188, 189, 195, 196 McCarthy 93, 94, 107, 110 McCloskey 113, 137 McGlone 258, 264 Merlini Barbaresi 35, 38, 39, 46, 54, 61, 63 Miller 30, 31, 229, 239, 242, 252, 263, 264 Mittwoch 43, 63 Montgomery 4, 11 Moon 260, 261, 264 Morley 25, 38 Myers 175, 197 N Nattinger 199, 214 Nystrand 200, 215 Nyyssönen 86, 91, 92, 110 O Ochs 118, 137, 138
Östman 4, 12, 156, 168, 171 P Park 153, 155 Peirce 19, 27, 28, 30, 38, 54, 63 Pennycook 209, 215 Poos 182, 195, 197 Prince 173, 174, 197, 229, 230, 238, 242 Q Quinn 258, 265 Quirk 2, 47, 61, 63, 65, 86, 120, 138, 219– 221, 240, 242 R Reinhart 43, 63, 155 Rey-Debove 21, 38 Rosenbaum 219, 242 Roth 118, 138 Russell 32, 38 S Sacks 4, 12, 242 Sadock 238, 242 Salager-Meyer 175, 197, 199, 215 Samson 6, 199, 206, 209, 215 Sarangi 11, 13 SchegloŸ 4, 12, 137, 138, 235, 242 SchiŸrin 4, 12, 119, 130, 138, 140, 141, 146, 153, 156, 158, 163–165, 167, 168, 171 Schober 90, 110 Schröder 173, 175, 196, 197, 215 Searle 3, 12, 42, 58, 63, 263, 265 Shroyer 50, 62 Silverstein 154, 156 Simon-Vandenbergen 3, 12, 254, 265 Simpson 138, 173, 175, 178, 182, 183, 195–197 Sinclair 2–5, 13, 91, 110, 140, 142, 156, 171, 176, 196, 197 Snell-Hornby 249, 265 Sperber 253, 265 Stampe 28, 36 Stati 146, 154, 156
270 Index of names
Steen, E. 71, 74, 86 Steen, G. 262, 263, 265 Stenström 162, 170, 171 Stern 254, 260, 265 Stubbs 1, 2, 4, 6, 13, 187, 197 Svartvik 2, 5, 13, 47, 61, 63, 86, 138, 242 Swales 10, 13, 138, 153, 156, 196, 197, 213, 216 T Takahara 153, 156 Tang 199, 213, 216 Tanskanen 6, 89, 93, 109, 110 Teich 80, 85, 86 Teubert 5, 13 Thesen 199, 216 Thompson, G. 140, 256 Thompson, S. 200, 216 Thompson, S.A. 7, 12. 110, 137, 138, 242 Tottie 5, 13 Traxler 90, 110 Tsui 4, 13, 158, 160–163, 171 Turner 43, 256, 264, 266 V van Dijk 2, 3, 24, 25, 38
Varttala 175, 195, 197 Verschueren 12, 62, 154, 156 Virtanen 4, 12, 65, 70, 77, 86 W Wales 203, 205, 216 Wallace 24, 38 Weinrich 21, 38 Werth 248, 258, 259, 265 Wikberg 10, 110, 245, 248, 260, 261, 263, 265 Wilkes-Gibbs 90, 109, 110 Wilson 11, 13, 55, 63, 253, 265 Winter 7, 13, 140, 156 Wodak 2, 3, 13 Y Young 199, 216 Yule 1–3, 11, 47, 62, 162, 171 Z Zacharski 48, 50, 63 Zhou 140, 156 Zimmer 33, 38 Ziv 140, 155, 156
Index of terms
A a little bit 174, 178, 190–195 abstracts 5, 9, 139–155 academic discourse 5, 6, 115, 129, 139, 152, 153, 175, 199, 201, 212 academic talk 173, 196 academic writing 7, 176, 199, 204, 225, 226, 239 anaphora 8, 19, 21, 43, 44, 55, 57, 61, 62, 119, 121, 130, 230 anaphoric reference 76 anticipatory it 235, 239 argumentation 141, 142, 153 B Bank of English 5, 174, 181, 192, 196 Bergen Corpus of London teenage language 5 Birmingham school 4 British National Corpus 5, 6, 8, 40, 92, 107, 115, 143, 261, 262 C cataphora 8, 21, 26, 30, 39–44, 46–49, 55– 57, 60 relational 43 sensu lato 41, 42, 48 sensu stricto 42 cataphoric reference 8, 19, 54, 60, 61 clefts 48–50, 55, 56, 221 Cobuild Corpus 5, 9 coherence 2, 8–10, 25, 26, 75, 89, 90, 96, 97, 140, 163, 164, 248 evaluative 9, 140 cohesion 2, 6, 7, 10, 25, 35, 42, 43, 47, 65, 67–74, 76, 77, 81–84, 89, 91–93, 97–99,
102, 104, 105, 106, 124, 137, 140, 234, 259, 262 cohesive devices 8, 90, 91, 100, 105, 106 lexical cohesion 10, 35, 43, 70–72, 76, 83, 84, 91, 92, 97, 98, 262 collaboration 6, 89, 90, 105–107 collocation 72, 76, 93, 95–98, 100–106, 143, 167, 262 activity-related 96, 98, 102, 104 elaborative 98, 102 comment 21, 73, 180, 232, 233, 235, 238, 239 complexity 8, 17, 19, 20, 26–29, 31, 33–36, 39, 40, 54–60, 227, 251, 253 concession 147, 153 conjunct 65, 67, 70–72, 74, 78, 82, 84, 146 connector 141, 142, 144, 146–149, 152, 153 contrastive connector 141, 152 contrastive rhetoric 10 Conversation Analysis 4, 114, 171 corpus linguistics 1 cultural determinism 92, 93 D deixis 47, 52, 114, 117–125, 127–130, 132, 135, 136 anaphoric 121 distal 121, 137 gestural 122, 127–129, 132 symbolic 122, 124, 128–130, 135 diagrammaticity 28, 29, 54, 57 discourse analysis 1–4, 10–13, 214, 262 discourse marker 140, 156 disjunct 67, 72, 73, 82, 85
272 Index of terms
E ellipsis 70, 71, 117 encapsulation 140 end-focus 232, 234 end-weight principle 223, 227, 228 English-Norwegian Parallel Corpus 10, 69 extraposition 7, 219–229, 232–235, 237– 240 F face-saving 159, 175, 194 Finnish 245, 248–250, 252, 253, 257 G general nouns 30, 35, 42, 48, 55, 56 German 10, 67, 69, 73, 79–84, 161, 245, 249 given-before-new principle 234, 237 H hedge 173, 176, 177, 182, 183, 188, 190– 195 however 9, 141, 143–153 I I don’t know 9, 45, 157–171, 185 iconicity 28, 30, 34–36, 54, 56, 57, 60, 61, 75 idiom 51, 92, 260 indexicality 17, 19, 29, 35, 40, 50, 54, 57 information packaging 229 information status 220, 229, 231, 232, 240 information structure 5, 60, 84, 229, 232, 234 it-extraposition 219, 220, 222, 224, 226, 228, 232, 233, 240, 241 J just 174,176, 186, 188–191, 193–195 K key words 5, 9 kind of 7, 174, 178–183, 194, 195
L Lancaster IBM Spoken English Corpus 97, 107 light subject constraint 7, 223, 227, 229, 237 London-Lund Corpus 5, 8, 40, 44, 47, 48, 51, 52, 61, 96, 107 M macro-connectivity 146, 147 markedness 8, 17, 19, 20, 28, 29, 34, 39, 40, 54, 56, 57, 60, 61, 220 metadiscourse 114, 140, 142, 153 interpersonal 142 metafunction 10, 68, 72, 73, 140, 262 textual 68, 86, 262 metaphor 10, 11, 19, 33, 120, 183, 245– 254, 256–260, 262 conceptual 257, 258 grammatical 258 Michigan Corpus of Academic Spoken English 5, 115, 173 mitigator 180, 190 modal particle 81, 82 N non-extraposition 7, 219–229, 232–235, 240 Norwegian 10, 67, 69, 80–85, 245, 246, 248, 250, 255 O or so 174, 176, 183, 185, 186, 194, 195 or something 174, 178, 183–186, 194, 195 Oslo Multilingual Corpus 10, 69, 246 P parallel corpora 5 person marker 202–204, 206, 208 phoricity 17, 18, 26, 30 point-making 141, 146, 149 pre-patterned speech 130 presupposition 238, 241
Index of terms 273
R recipient design 114, 126, 131 reiteration 72, 93, 95, 97–103, 105 Relevance Theory 253 repetition 93, 98, 99, 101–105, 108, 116, 121, 148, 206, 208, 230, 234 complex 93, 103 retrievability span 229 rhetorical relations 7, 8, 22 right-dislocation 220 S Santa Barbara Corpus of spoken American English 5 self-mention 201, 209, 210, 212 self-repair 235 Siena Corpus 5, 9, 115, 132, 133 simile 245, 249, 253, 255–257, 260 somewhat 176, 178, 186–191, 193–195 sort of 174, 177–179, 181–183, 186, 189, 194, 195 speciªcation 95, 98, 99, 101, 102, 104, 147, 186 co-speciªcation 95, 98, 99, 102 substitution 70, 71, 83, 93, 99, 100, 103, 104 synonymy 26, 94, 219, 239 systemic functional grammar 7, 86 T text-copying 97, 104, 105 texture 47, 65, 68, 69, 84
thematic structure 25, 68, 69, 82 theme 10, 22, 24, 33, 65–86, 148, 248, 255 conjunctive 74 experiential 66, 72, 73, 75, 81, 82, 85 interpersonal 67, 74–76, 82, 148 multiple 10, 66–68, 70, 72, 74, 76–78, 82–84 simple 72 textual 67, 82 topical 68, 80, 81 titles 8, 17–32, 34, 35 opaque 32 transparent 29–31, 34 topic 21, 25, 96, 97, 121, 140, 199, 205, 208, 211, 235, 237, 238 , 252, 253, 256, 260, 262 discourse topic 77, 229, 233, 236, 259 new topic 77, 79, 236, 239 topic break 10 topic change 8, 52 topic-comment structure 232 translation corpora 5, 248 V verb-second constraint 80, 84 visuals 115–118, 132–135 W Wh-clefts 221 Wordsmith tools 143, 144, 201
In the PRAGMATICS AND BEYOND NEW SERIES the following titles have been published thus far or are scheduled for publication: 1. WALTER, Bettyruth: The Jury Summation as Speech Genre: An Ethnographic Study of What it Means to Those who Use it. Amsterdam/Philadelphia, 1988. 2. BARTON, Ellen: Nonsentential Constituents: A Theory of Grammatical Structure and Pragmatic Interpretation. Amsterdam/Philadelphia, 1990. 3. OLEKSY, Wieslaw (ed.): Contrastive Pragmatics. Amsterdam/Philadelphia, 1989. 4. RAFFLER-ENGEL, Walburga von (ed.): Doctor-Patient Interaction. Amsterdam/Philadelphia, 1989. 5. THELIN, Nils B. (ed.): Verbal Aspect in Discourse. Amsterdam/Philadelphia, 1990. 6. VERSCHUEREN, Jef (ed.): Selected Papers from the 1987 International Pragmatics Conference. Vol. I: Pragmatics at Issue. Vol. II: Levels of Linguistic Adaptation. Vol. III: The Pragmatics of Intercultural and International Communication (ed. with Jan Blommaert). Amsterdam/Philadelphia, 1991. 7. LINDENFELD, Jacqueline: Speech and Sociability at French Urban Market Places. Amsterdam/Philadelphia, 1990. 8. YOUNG, Lynne: Language as Behaviour, Language as Code: A Study of Academic English. Amsterdam/Philadelphia, 1990. 9. LUKE, Kang-Kwong: Utterance Particles in Cantonese Conversation. Amsterdam/Philadelphia, 1990. 10. MURRAY, Denise E.: Conversation for Action. The computer terminal as medium of communication. Amsterdam/Philadelphia, 1991. 11. LUONG, Hy V.: Discursive Practices and Linguistic Meanings. The Vietnamese system of person reference. Amsterdam/Philadelphia, 1990. 12. ABRAHAM, Werner (ed.): Discourse Particles. Descriptive and theoretical investigations on the logical, syntactic and pragmatic properties of discourse particles in German. Amsterdam/Philadelphia, 1991. 13. NUYTS, Jan, A. Machtelt BOLKESTEIN and Co VET (eds): Layers and Levels of Representation in Language Theory: A functional view. Amsterdam/Philadelphia, 1990. 14. SCHWARTZ, Ursula: Young Children’s Dyadic Pretend Play. Amsterdam/Philadelphia, 1991. 15. KOMTER, Martha: Conflict and Cooperation in Job Interviews. Amsterdam/Philadelphia, 1991. 16. MANN, William C. and Sandra A. THOMPSON (eds): Discourse Description: Diverse Linguistic Analyses of a Fund-Raising Text. Amsterdam/Philadelphia, 1992. 17. PIÉRAUT-LE BONNIEC, Gilberte and Marlene DOLITSKY (eds): Language Bases ... Discourse Bases. Amsterdam/Philadelphia, 1991. 18. JOHNSTONE, Barbara: Repetition in Arabic Discourse. Paradigms, syntagms and the ecology of language. Amsterdam/Philadelphia, 1991. 19. BAKER, Carolyn D. and Allan LUKE (eds): Towards a Critical Sociology of Reading Pedagogy. Papers of the XII World Congress on Reading. Amsterdam/Philadelphia, 1991. 20. NUYTS, Jan: Aspects of a Cognitive-Pragmatic Theory of Language. On cognition, functionalism, and grammar. Amsterdam/Philadelphia, 1992. 21. SEARLE, John R. et al.: (On) Searle on Conversation. Compiled and introduced by Herman Parret and Jef Verschueren. Amsterdam/Philadelphia, 1992. 22. AUER, Peter and Aldo Di LUZIO (eds): The Contextualization of Language. Amsterdam/ Philadelphia, 1992.
23. FORTESCUE, Michael, Peter HARDER and Lars KRISTOFFERSEN (eds): Layered Structure and Reference in a Functional Perspective. Papers from the Functional Grammar Conference, Copenhagen, 1990. Amsterdam/Philadelphia, 1992. 24. MAYNARD, Senko K.: Discourse Modality: Subjectivity, Emotion and Voice in the Japanese Language. Amsterdam/Philadelphia, 1993. 25. COUPER-KUHLEN, Elizabeth: English Speech Rhythm. Form and function in everyday verbal interaction. Amsterdam/Philadelphia, 1993. 26. STYGALL, Gail: Trial Language. A study in differential discourse processing. Amsterdam/ Philadelphia, 1994. 27. SUTER, Hans Jürg: The Wedding Report: A Prototypical Approach to the Study of Traditional Text Types. Amsterdam/Philadelphia, 1993. 28. VAN DE WALLE, Lieve: Pragmatics and Classical Sanskrit. Amsterdam/Philadelphia, 1993. 29. BARSKY, Robert F.: Constructing a Productive Other: Discourse theory and the convention refugee hearing. Amsterdam/Philadelphia, 1994. 30. WORTHAM, Stanton E.F.: Acting Out Participant Examples in the Classroom. Amsterdam/Philadelphia, 1994. 31. WILDGEN, Wolfgang: Process, Image and Meaning. A realistic model of the meanings of sentences and narrative texts. Amsterdam/Philadelphia, 1994. 32. SHIBATANI, Masayoshi and Sandra A. THOMPSON (eds): Essays in Semantics and Pragmatics. Amsterdam/Philadelphia, 1995. 33. GOOSSENS, Louis, Paul PAUWELS, Brygida RUDZKA-OSTYN, Anne-Marie SIMONVANDENBERGEN and Johan VANPARYS: By Word of Mouth. Metaphor, metonymy and linguistic action in a cognitive perspective. Amsterdam/Philadelphia, 1995. 34. BARBE, Katharina: Irony in Context. Amsterdam/Philadelphia, 1995. 35. JUCKER, Andreas H. (ed.): Historical Pragmatics. Pragmatic developments in the history of English. Amsterdam/Philadelphia, 1995. 36. CHILTON, Paul, Mikhail V. ILYIN and Jacob MEY: Political Discourse in Transition in Eastern and Western Europe (1989-1991). Amsterdam/Philadelphia, 1998. 37. CARSTON, Robyn and Seiji UCHIDA (eds): Relevance Theory. Applications and implications. Amsterdam/Philadelphia, 1998. 38. FRETHEIM, Thorstein and Jeanette K. GUNDEL (eds): Reference and Referent Accessibility. Amsterdam/Philadelphia, 1996. 39. HERRING, Susan (ed.): Computer-Mediated Communication. Linguistic, social, and crosscultural perspectives. Amsterdam/Philadelphia, 1996. 40. DIAMOND, Julie: Status and Power in Verbal Interaction. A study of discourse in a closeknit social network. Amsterdam/Philadelphia, 1996. 41. VENTOLA, Eija and Anna MAURANEN, (eds): Academic Writing. Intercultural and textual issues. Amsterdam/Philadelphia, 1996. 42. WODAK, Ruth and Helga KOTTHOFF (eds): Communicating Gender in Context. Amsterdam/Philadelphia, 1997. 43. JANSSEN, Theo A.J.M. and Wim van der WURFF (eds): Reported Speech. Forms and functions of the verb. Amsterdam/Philadelphia, 1996. 44. BARGIELA-CHIAPPINI, Francesca and Sandra J. HARRIS: Managing Language. The discourse of corporate meetings. Amsterdam/Philadelphia, 1997. 45. PALTRIDGE, Brian: Genre, Frames and Writing in Research Settings. Amsterdam/ Philadelphia, 1997.
46. GEORGAKOPOULOU, Alexandra: Narrative Performances. A study of Modern Greek storytelling. Amsterdam/Philadelphia, 1997. 47. CHESTERMAN, Andrew: Contrastive Functional Analysis. Amsterdam/Philadelphia, 1998. 48. KAMIO, Akio: Territory of Information. Amsterdam/Philadelphia, 1997. 49. KURZON, Dennis: Discourse of Silence. Amsterdam/Philadelphia, 1998. 50. GRENOBLE, Lenore: Deixis and Information Packaging in Russian Discourse. Amsterdam/Philadelphia, 1998. 51. BOULIMA, Jamila: Negotiated Interaction in Target Language Classroom Discourse. Amsterdam/Philadelphia, 1999. 52. GILLIS, Steven and Annick DE HOUWER (eds): The Acquisition of Dutch. Amsterdam/ Philadelphia, 1998. 53. MOSEGAARD HANSEN, Maj-Britt: The Function of Discourse Particles. A study with special reference to spoken standard French. Amsterdam/Philadelphia, 1998. 54. HYLAND, Ken: Hedging in Scientific Research Articles. Amsterdam/Philadelphia, 1998. 55. ALLWOOD, Jens and Peter Gärdenfors (eds): Cognitive Semantics. Meaning and cognition. Amsterdam/Philadelphia, 1999. 56. TANAKA, Hiroko: Language, Culture and Social Interaction. Turn-taking in Japanese and Anglo-American English. Amsterdam/Philadelphia, 1999. 57 JUCKER, Andreas H. and Yael ZIV (eds): Discourse Markers. Descriptions and theory. Amsterdam/Philadelphia, 1998. 58. ROUCHOTA, Villy and Andreas H. JUCKER (eds): Current Issues in Relevance Theory. Amsterdam/Philadelphia, 1998. 59. KAMIO, Akio and Ken-ichi TAKAMI (eds): Function and Structure. In honor of Susumu Kuno. 1999. 60. JACOBS, Geert: Preformulating the News. An analysis of the metapragmatics of press releases. 1999. 61. MILLS, Margaret H. (ed.): Slavic Gender Linguistics. 1999. 62. TZANNE, Angeliki: Talking at Cross-Purposes. The dynamics of miscommunication. 2000. 63. BUBLITZ, Wolfram, Uta LENK and Eija VENTOLA (eds.): Coherence in Spoken and Written Discourse. How to create it and how to describe it.Selected papers from the International Workshop on Coherence, Augsburg, 24-27 April 1997. 1999. 64. SVENNEVIG, Jan: Getting Acquainted in Conversation. A study of initial interactions. 1999. 65. COOREN, François: The Organizing Dimension of Communication. 2000. 66. JUCKER, Andreas H., Gerd FRITZ and Franz LEBSANFT (eds.): Historical Dialogue Analysis. 1999. 67. TAAVITSAINEN, Irma, Gunnel MELCHERS and Päivi PAHTA (eds.): Dimensions of Writing in Nonstandard English. 1999. 68. ARNOVICK, Leslie: Diachronic Pragmatics. Seven case studies in English illocutionary development. 1999. 69. NOH, Eun-Ju: The Semantics and Pragmatics of Metarepresentation in English. A relevance-theoretic account. 2000. 70. SORJONEN, Marja-Leena: Responding in Conversation. A study of response particles in Finnish. 2001. 71. GÓMEZ-GONZÁLEZ, María Ángeles: The Theme-Topic Interface. Evidence from English. 2001.
72. MARMARIDOU, Sophia S.A.: Pragmatic Meaning and Cognition. 2000. 73. HESTER, Stephen and David FRANCIS (eds.): Local Educational Order. Ethnomethodological studies of knowledge in action. 2000. 74. TROSBORG, Anna (ed.): Analysing Professional Genres. 2000. 75. PILKINGTON, Adrian: Poetic Effects. A relevance theory perspective. 2000. 76. MATSUI, Tomoko: Bridging and Relevance. 2000. 77. VANDERVEKEN, Daniel and Susumu KUBO (eds.): Essays in Speech Act Theory. 2002. 78. SELL, Roger D. : Literature as Communication. The foundations of mediating criticism. 2000. 79. ANDERSEN, Gisle and Thorstein FRETHEIM (eds.): Pragmatic Markers and Propositional Attitude. 2000. 80. UNGERER, Friedrich (ed.): English Media Texts – Past and Present. Language and textual structure. 2000. 81. DI LUZIO, Aldo, Susanne GÜNTHNER and Franca ORLETTI (eds.): Culture in Communication. Analyses of intercultural situations. 2001. 82. KHALIL, Esam N.: Grounding in English and Arabic News Discourse. 2000. 83. MÁRQUEZ REITER, Rosina: Linguistic Politeness in Britain and Uruguay. A contrastive study of requests and apologies. 2000. 84. ANDERSEN, Gisle: Pragmatic Markers and Sociolinguistic Variation. A relevance-theoretic approach to the language of adolescents. 2001. 85. COLLINS, Daniel E.: Reanimated Voices. Speech reporting in a historical-pragmatic perspective. 2001. 86. IFANTIDOU, Elly: Evidentials and Relevance. 2001. 87. MUSHIN, Ilana: Evidentiality and Epistemological Stance. Narrative retelling. 2001. 88. BAYRAKTAROG LU, ArFn and Maria SIFIANOU (eds.): Linguistic Politeness Across Boundaries. The case of Greek and Turkish. 2001. 89. ITAKURA, Hiroko: Conversational Dominance and Gender. A study of Japanese speakers in first and second language contexts. 2001. 90. KENESEI, István and Robert M. HARNISH (eds.): Perspectives on Semantics, Pragmatics, and Discourse. A Festschrift for Ferenc Kiefer. 2001. 91. GROSS, Joan: Speaking in Other Voices. An ethnography of Walloon puppet theaters. 2001. 92. GARDNER, Rod: When Listeners Talk. Response tokens and listener stance. 2001. 93. BARON, Bettina and Helga KOTTHOFF (eds.): Gender in Interaction. Perspectives on femininity and masculinity in ethnography and discourse. 2002 94. McILVENNY, Paul (ed.): Talking Gender and Sexuality. 2002. 95. FITZMAURICE, Susan M.: The Familiar Letter in Early Modern English. A pragmatic approach. 2002. 96. HAVERKATE, Henk: The Syntax, Semantics and Pragmatics of Spanish Mood. 2002. 97. MAYNARD, Senko K.: Linguistic Emotivity. Centrality of place, the topic-comment dynamic, and an ideology of pathos in Japanese discourse. 2002. 98. DUSZAK, Anna (ed.): Us and Others. Social identities across languages, discourses and cultures. 2002. 99. JASZCZOLT, K.M. and Ken TURNER (eds.): Meaning Through Language Contrast. Volume 1. 2003. 100. JASZCZOLT, K.M. and Ken TURNER (eds.): Meaning Through Language Contrast. Volume 2. 2003.
101. LUKE, Kang Kwong and Theodossia-Soula PAVLIDOU (eds.): Telephone Calls. Unity and diversity in conversational structure across languages and cultures. 2002. 102. LEAFGREN, John: Degrees of Explicitness. Information structure and the packaging of Bulgarian subjects and objects. 2002. 103. FETZER, Anita and Christiane MEIERKORD (eds.): Rethinking Sequentiality. Linguistics meets conversational interaction. 2002. 104. BEECHING, Kate: Gender, Politeness and Pragmatic Particles in French. 2002. 105. BLACKWELL, Sarah E.: Implicatures in Discourse. The case of Spanish NP anaphora. 2003. 106. BUSSE, Ulrich: Linguistic Variation in the Shakespeare Corpus. Morpho-syntactic variability of second person pronouns. 2002. 107. TAAVITSAINEN, Irma and Andreas H. JUCKER (eds.): Diachronic Perspectives on Address Term Systems. 2003. 108. BARRON, Anne: Acquisition in Interlanguage Pragmatics. Learning how to do things with words in a study abroad context. 2003. 109. MAYES, Patricia: Language, Social Structure, and Culture. A genre analysis of cooking classes in Japan and America. 2003. 110. ANDROUTSOPOULOS, Jannis K. and Alexandra GEORGAKOPOULOU (eds.): Discourse Constructions of Youth Identities. 2003. 111. ENSINK, Titus and Christoph SAUER (eds.): Framing and Perspectivising in Discourse. 2003. 112. LENZ, Friedrich (ed.): Deictic Conceptualisation of Space, Time and Person. 2003. 113. PANTHER, Klaus-Uwe and Linda L. THORNBURG (eds.): Metonymy and Pragmatic Inferencing. 2003. 114. KÜHNLEIN, Peter, Hannes RIESER and Henk ZEEVAT (eds.): Perspectives on Dialogue in the New Millennium. 2003. 115. KÄRKKÄINEN, Elise: Epistemic Stance in English Conversation. A description of its interactional functions, with a focus on I think. 2003. 116. GRANT, Colin B. (ed.): Rethinking Communicative Interaction. New interdisciplinary horizons. 2003. 117. WU, Ruey-Jiuan Regina: Stance in Talk. A conversation analysis of Mandarin final particles. n.y.p. 118. CHENG, Winnie: Intercultural Conversation. 2003. 119. HILTUNEN, Risto and Janne SKAFFARI (eds.): Discourse Perspectives on English. Medieval to modern. 2003. 120. AIJMER, Karin and Anna-Brita STENSTRÖM (eds.): Discourse Patterns in Spoken and Written Corpora. 121. FETZER, Anita: Recontextualizing Context. Grammaticality meets appropriateness. 2004. 122. GONZÁLEZ, Montserrat: Pragmatic Markers in Oral Narrative. The case of English and Catalan. n.y.p. 123. MÁRQUEZ REITER, Rosina and María Elena PLACENCIA (eds.): Current Trends in the Pragmatics of Spanish. n.y.p. 124. VINE, Bernadette: Getting Things Done at Work. The discourse of power in workplace interaction. n.y.p. 125. LERNER, Gene H. (ed.): Conversation Analysis. Studies from the first generation. n.y.p. 126. WU, Yi’an: Spatial Demonstratives in English and Chinese. Text and Cognition. n.y.p.
