Grammatical Metaphor / Metonymy in the Treaty Establishing A constitution for Europe : A Comparison Between the English and Swedish Versions

Grammatical metaphor is attracting increasing interest, together with a stronger focus on non-literary texts as the subjects of analysis—text types where grammatical metaphor is most commonly used. Grammatical metaphor may be used in both English and Swedish, but seems to be considerably less frequent in Swedish. A comparison of similar texts may indicate restrictions in linguistic usage and also contribute to long-term studies of what may become an example of linguistic change under the influence of legally imposed translation of legislative texts.


Introduction
Grammatical metaphor is attracting increasing interest, together with a stronger focus on non-literary texts as the subjects of analysis-text types where grammatical metaphor is most commonly used.Grammatical metaphor may be used in both English and Swedish, but seems to be considerably less frequent in Swedish.A comparison of similar texts may indicate restrictions in linguistic usage and also contribute to long-term studies of what may become an example of linguistic change under the influence of legally imposed translation of legislative texts.

Background
The growing interest in metaphor is connected with the equally rapidly growing attention surrounding non-literary texts.Thus the previously dominant study of metaphor as a stylistic trope found in literary texts has now been largely replaced by investigations of other kinds of metaphoric language, such as Lakoffian studies of cognitive, or conceptual metaphor, or the use of grammatical metaphor, in a wide variety of texts where fiction only figures marginally.
The current interest in non-literary texts may be inspired by different needs, the increasing global need for English for special purposes being one, the growing awareness of the social impact of textual strategies, another.Whatever the linguists' reasons may be, the fact remains that such texts constitute the overwhelming majority of all printed matter, a reason as good as any for devoting more time and effort to analysing them.Although a large proportion of such texts are produced by and for specialists, the very importance of their subject matter-be it technical, scientific, legal, financial or political-in combination with the prestige of many of their authors creates a trickle-down effect that influences language for general purposes, i.e. language at large.Similar interference may be noted between language users of different languages connected by similar professions, disciplines or areas of interest: as shown in a comparison between Swedish and British financial journalists (Stålhammar 1995), similar metaphorical expressions were used by both British and Swedish writers.Lexical loans have long been noted, but what has attracted less attention is the less visible impact on the grammar of a language or at the receiving end of cultural influence, for example the target language (TL) in translation.

2. Grammatical metaphor versus metonymy
If semantic metaphor is defined as the substitution of one word by another, grammatical metaphor may be defined as "a substitution of one grammatical class, or one grammatical structure, by another" (Halliday & Martin 1993:79).The term "grammatical metaphor" was introduced by Halliday (cf Halliday 1985/1994, Halliday & Martin 1993) and fulfils certain important needs: in Halliday's words, it "opens up a new dimension of semantic space" (Halliday 2003).The study of the uses of grammatical metaphor is particularly useful in revealing how processes are reified into objects, thus altering not only the grammar of texts but also reader reactions to texts.
Being a form of condensation of information (process + actor), grammatical metaphor is a very economical means of packaging information and is consequently frequently used in scientific and technical information.In these areas, the "real" actors are often absent from the scene, replaced by the (nominalised) processes they have set in motion.Grammatical metaphor thus serves more than the purpose of efficiency: it lends an appearance of objectivity and anonymity to the text.The association of grammatical metaphor with scientific and technical text types may have contributed to negative reactions against the "reification" implicit in the construction (cf Goatly's discussion of underlying political assumptions, Goatly 1996), or simply against the removal of (human) agency, and consequently responsibility, from texts.
In the following, grammatical metaphor will be the only term borrowed from Systemic Functional Grammar and no attempt will be made to apply other features from SFG.
There has so far been little discussion of possible differences between various types of grammatical metaphor.Halliday's examples are clear instances of condensed processes and later research has followed the same lines of investigation, in accordance with Halliday's view that "nominalizing is the single most powerful resource for creating grammatical metaphor" (Halliday 1994:353).Grammatical metaphor in commonly cited examples tends to involve nominalisation with traceable congruent actors/ processes, where unpackaging into underlying clauses may require both effort, contextual knowledge and linguistic maturity.In all these respects, grammatical metaphor has similarities with "ordinary" metaphor.The requirements necessary for processing grammatical metaphor have also been studied both for adult language users (Lassen 2003) and for language learners (Halliday 1993, Gibbons 1998, Mohan & Beckett 2001, Derewianka 2003) and results indicate that grammatical metaphor requires a certain level of linguistic competence.
Whereas metaphor is defined as the transfer of meaning between two dissociated concepts, metonymy, relying on contiguity, lacks this aspect of difference, or tension, between the concepts involved.Metonymy is based on close association, for example designation of x by some element associated with x, for example agent for action pen for writing , producer for product (Dickens for Dickens's works); instrument for user (cello for cellist); container for contents (the Houses of Parliament for the Members of Parliament, or parliamentary work).Synecdoche, part for whole or vice versa (hands for workers, The Guardian for a reporter from The Guardian), has strong similarities with metonymy and is here subsumed under this term.
The distinction between both concepts is far from clear-cut and is the subject of increasing discussion (cf Jakobson and others in the anthologies edited by Barcelona 2000 andDirven &Pörings 2002), although studies of metaphor have dominated since the early days of rhetoric.Metonymy is often considered as being of less interest because of less originality: its very essence is predictability (the association must be obvious), whereas metaphor is traditionally praised for its originality, its heuristic effect of showing hitherto unnoticed connections.
The concept of grammatical metaphor (GM) includes, however, a number of concepts where unpacking seems to serve little purpose, and where there is considerably less "tension" than in nominalised processes.Such examples include terms for abstract phenomena like study, paper, investigation, etc, i.e. obvious metonymic substitutes for the authors of these works.Similarly, various terms for the results of political and legal processes, like law, court, institution can be seen as metonymies replacing the anonymous actors participating in the processes.
There may thus be considerable advantages to be gained from separating grammatical metaphor, with its high processing requirements and depersonalising effects, from metonymical cases.(For a hierarchical representation of different categories of grammatical metaphor along a lexico-grammatical scale, see Lassen 1997:75-76.)Research into language development points towards metonymy appearing before metaphor in child language development (Nerlich et al 1999), which indicates that there may be less difficulty associated with processing metonymy than metaphor.If this is confirmed by further research, it may be another argument in favour of such a separation.
In the present study, I will tentatively propose grammatical metonymy (GMy) for such cases where a metonymic relationship is obvious between the explicit (incongruent) subject and an implied (congruent) one.

Previous research
As indicated above, studies related to grammatical metaphor have grown rapidly in number since the introduction of the concept in Halliday's first edition of his Introduction to Functional Grammar (1986).The interest in the use of grammatical metaphor in specialised language use seems to have increased after Halliday's and Martin's analysis of the use of grammatical metaphor in technical and scientific writing, in their Writing Science: Literacy and Discursive Power (1993).Among studies relevant to the present investigation are Lassen (2003), Banks (2003).For an annotated bibliography covering publications up to 2002, see Taverniers 'list <http://bank.rug.ac.be/mt/gm/gmbiblio.html>.
Except for the large body of Chinese studies (cf Huang 2002), studies of grammatical metaphor in other languages than English are less frequent (cf Maagerø 1997, Gibbons and Lascar 1998, Plementitas 1998).
Among studies of language development, including the use of grammatical metaphor, Derewianka's (2003) study is the most relevant, being among the few to proceed into adolescence and include instances similar to those under investigation in the present study.The use of metonymy among small children is the subject of a case study by Nerlich et al (1999); Halliday's studies of various aspects of children's language 1969-1999 have been collected in one volume (Halliday 1999).

Aim
The aim of the present study is to investigate the use of a particular type of grammatical metaphor, here called grammatical metonymy, in two different versions of a central, contemporary political document, the proposed constitution of the European Union.The usage under study is frequent in English, here termed the source language (SL) and consequently in the source text (ST).The Swedish version, here termed the target text (TT) was compared for those passages where ST uses grammatical metaphor.(These terms will be used throughout, despite the EU definition of all language versions of central treaties as being "equally authentic" (Article IV-448), a definition that consequently excludes the very notion of translation.)Recurring patterns were investigated for common features leading to certain translation strategies.
The material of the present study, the English and Swedish versions of the proposed constitution of the European Union, fulfils the criteria mentioned above: such texts will be widely read and quoted and may thus influence the language of the member country, both through their actual legal status and through their prestige.The language used may consequently contribute to the development of the national language.Even if the present text of the constitution is not adopted, the language use in the Swedish version may bear witness to the translators' different choices and strategies.An analysis of such a central document may thus be of both diachronic and synchronic interest.

Material and method
The material consists of a legal text intended as the constitution for the European Union.After negative votes from the French and Dutch electorates, the text was withdrawn in 2005.The present analysis was continued, however, and may serve as a comparison with possible later versions.

Material
The material consists of the electronic and paper versions of the Treaty establishing a Constitution for Europe in English and Swedish, here called EUC-E and EUC-S, respectively.Both versions are remarkably similar in size: preambles and other introductory material excluded, the printed version of EUC-E numbers 181 pages and that of EUC-S 182 pages, or, in characters (including spaces): EUC-E 416,421 and EUC-S 403,612; the word count resulted in EUC-E 66,322 and EUC-S 56,535 words, respectively.Using characters instead of words for quantitative comparisons has the advantage of avoiding language specific characteristics such as the formation of solid or open compounds (with or without particles).It may be noted that all these figures contradict the nearly unanimous notions that translations inevitably exceed originals in length.
For comparison with Swedish language usage, the concordances in the Svenska Språkbanken at Göteborg University were used (approximately 100 million words of newspaper text).

Method
The material not being tagged, manual search was the only reliable search method.All grammatical metaphors defined as above in the ST were excerpted and in an attempt to safeguard against omissions, all instances found were electronically checked, which led to a small number of further occurrences.Once these searches were carried out, the TT was excerpted for all the corresponding items, a total of 163 excerpts in either text.
A comparison with a previous study of a similar kind (Stålhammar 2003) resulted in 18 items in a considerably shorter set of STs, the Maastricht Treaty in English, Swedish, French and German, cf Stålhammar 2003.(Counted in characters, the text amounts to: English 52,998; Swedish 51,044; French 55,379, German 55,430, respectively.The size of the document measured in number of pages in the printed version is difficult to compare, owing to the parallel publication of four languages in columns side by side.)These figures indicate a rather similar distribution of grammatical metaphors in the different treaties, with approximately nine times as many occurrences in a roughly eight times larger material.

Results
To illustrate the most commonly recurring patterns, some examples will be given with back-translations of the Swedish version in square brackets.

English active verb -Swedish active verb:
Eng "European laws shall contribute" Sw "Europeiska lagar skall bidra" [European laws shall contribute] English active verb -Swedish passive with adverbial: Eng "European law shall establish programmes" Sw "program skall fastställas i europeiska lagar" [programmes shall be established in European law] English active verb -Swedish nominalisation: Eng "such laws shall aim to" Sw " syftet med dessa lagar" [the aim of such laws] English active verb -Swedish vara [be]with predicate: Eng "acts shall bind states" Sw "akter skall vara bindande för stater" [acts shall be binding for states] English active verb -Swedish subject change and adverbial: Eng "a European law of the Council may establish [an] Office" Sw "får rådet genom en europeisk lag inrätta en ... myndighet" [the Council may, by a European law, establish an office] The last example shows clearly the metonymic relationship between legislator and law.

Translations of grammatical metaphor in EUC-E into EUC-S
Quantitatively, the use of the passive (with adverbial) dominates strongly, with approximately three times as many instances (119 compared to 40).Very few other solutions were found, e.g.other nominalisation (3), subject change (1), or be with predicate (1).For details, see Table 1.
A comparison with the previous comparative study of different versions of the Maastricht treaty shows a different result: here active verbs outnumber passives by two to one (11 active verbs, 5 passives, 2 be with predicate).Swedish translators seem thus to have become more cautious, in spite of internal discussions in favour of transferring active verb forms into Swedish translations (personal communication with Swedish EU language coordinator Kenneth Larsson 2005).
What restrictions may regulate the choice between active and passive verb forms?Swedish has a long history of using certain active verb forms, particularly reporting verbs, together with certain legal concepts: "lagen säger" [the law says] is a frequent expression that may be traced back to the Bible.Similarly, "lagen tillåter inte" [the law does not allow], and "lagen kräver" [the law requires] occur frequently in Språkbanken's corpus.At a lower level of legal nomenclature, paragraphs may also "say" something, although much less frequently, and acts and guidelines, (terms of much lower frequency) are not found with any active verbs.A certain legal hierarchy seems to apply in general language usage (a theory also proposed by Kenneth Larsson), and was consequently investigated in a more detailed analysis of the material, cf Table 1.As seen from Table 1, common language usage is not parallelled in this material: the term most often connected with active verbs in general language use, law, very rarely occurs with active verb forms, whereas legal concepts at both higher or equal levels of legal nomenclature (constitution, act, charter) and lower levels (article, chapter, decision, provision, subsection) seem to be more freely combined with active verbs, although within certain limits.The most frequently used non-stative active verb, påverka [affect] is primarily used to describe forms of abstract influence (the Swedish counterpart to the OED, the SAOB, almost exclusively exemplifies such usage), which may contribute to congruency in this context.
It is notable that the translators have deviated from the original in their efforts to avoid active verbs: by choosing to translate "the charter does not extend..." as "stadga[n] innebär inte någon utvidgning" [the charter does not signify any extension], the translator presents an interpretation of the content of the text, rather than a transfer of what is said in the text.A translation of "law/s /the programme shall establish", or "determine" as "lag/ar /programmet skall innehålla" [law/s /the programme shall contain] alters the performativity expressed in the legal source text.It could be argued that the performativity in the EUC-E examples is weakened in all instances of the EUC-S translations where active verbs are replaced by adverbials and passives.
The remaining incongruent verbs are few in number but the more surprising: kräva [require, demand] and respektera [respect].In the first case, the conventional Swedish construction with an adverbial and a passive verb would have been preferable: as it stands we have "guidelines and projects" presenting demands in a manner not generally seen in Swedish, where the fixed expression is exemplified in another paragraph: "för verkställighet krävs samtycke" [for implementation agreement is required].The translation of comply with by respektera [respect] is equally unidiomatic, since this verb in Swedish is exclusively connected with animate beings capable of understanding the notion of respect.
As shown by these examples, a cline of congruency/acceptance may be observed.It will be interesting to note possible changes in usage over time.The examples also show that the English version has a considerably smaller proportion of stative verbs, and a correspondingly larger share of verbs denoting activity.In other words, the English language seems to accept more different varieties of activity in abstract concepts.
The impression of almost mechanical, by-default translation of a large number of various active verbs into combinations of adverbials and passive verb forms could be explained by the use of so-called translation memories (TM), i.e. computerised segments of texts with corresponding translations.This possibility was excluded by information from Kenneth Larsson that the entire document (like the Maastricht treaty) was translated without TM support.

Context: Swedish stylistic recommendations and usage
In this context it is of particular interest to compare the Swedish version of a central text in the European Union with general recommendations for Swedish official texts.

Swedish official recommendations
The Swedish government has devoted special resources to language work with the laudable intention to simplify bureaucratic language in order to strengthen democracy.The government's home page, http:// www.regeringen.se,has links to various documents produced by "Klarspråksgruppen" the committee for plain language, advising Swedish citizens how to write simply and legibly, an effort similar to the Fight the Fog campaign in the European Union, or the Plain English movement in English-speaking countries.Added to these efforts, much of school and university training in writing focuses on similar recommendations: use the active voice, avoid the passive ("Att översätta EU-rättsakter", the Swedish guide to translating EU legislation advises translators to prefer constructions with man [one] rather than the passive, p.9), avoid long sentences (cf the government publication Klarspråksbulletinen 2005:4, p.6; very similar rules for Norwegian are described in Maagerø 1997).Translators and lawyers (including the specialised legal translators (juristlingvister) employed by Sweden's translation unit in the EU) are familiar with these recommendations, so the reader would expect Swedish EU texts produced along these guidelines.The sentence length cannot be altered, owing to European Union rules for legal documents, but sentence constructions could certainly be improved according to government recommendations.

Passive verbs in the EUC-S
In stark contrast to the recommendations described above, the Swedish version of the constitution is dominated by passive verb forms.This recurrent pattern is further aggravated by the voluminous premodifiers that are turned into passive clauses, thus creating double passives.The recurring (24 instances) phrases "European law /or framework law of the Council", is only once translated as "rådets europeiska lag", but in all other examples translated as "en europeisk lag som antas av rådet" [a European law which is passed by the Council], i.e. by the addition of a relative clause with a passive verb.This entails that a simple construction like the English "a European law or framework law of the Council may establish measures concerning passports" becomes the Swedish "får det i en europeisk lag eller ramlag som antas av rådet föreskrivas åtgärder om pass" (passive forms italicised) [measures may be established concerning passports in a European law or a law which has been passed by the Council], i.e. a sentence involving two examples of the abhorred passive.

Active verbs in Swedish general language usage
A comparison with only three years' material (2002)(2003)(2004)  These more than 30 different active verbs combined with lag [law] in general Swedish language use are evidence that there is acceptance for this type of incongruent use.There should thus be options other than the passive with/without adverbials for translations of corresponding verbs from English.

Metonymy or metaphor?
Of the 20 subject types found in the source text material, by far the largest category are typical metonymies of the act-for-actor type (law for legislator), viz (in alphabetic order) act, article, chapter, charter, constitution, cooperation (here denoting organised cooperation projects), decision, framework (in this context a legal category), guideline, law, policy, programme, proposal, provision, report, subsection.Fund and institution may be seen as collective labels for members of an organisation and thus as examples of metonymic relationships.The few remaining ones, action, implementation, are closer to conventional (albeit rather unclearly defined) grammatical metaphors.Grammatical metonymy, easier to process, thus outnumbers grammatical metaphor/nominalisation commonly associated with specialised writing, a fact which may be a result of a wish to write accessibly for the common citizen.
As shown above (Tables 1, 2), active verbs occurred with almost all abstract concepts in subject position in the source text, EUC-E.In the target text, EUC-S, no clear patterns emerge for any particular subject, but the passive dominates strongly, with 119 instances compared to 44 active ones for the 168 source text examples (5 were translated by other solutions), i.e. active verbs are used in approximately one fourth of all cases.
Excluding the translations of law + active verb, however, alters the pattern: the 50 EUC-E examples (tokens) with other subjects than law are translated by active verbs in 31 cases, i.e. more than half.It thus emerges, somewhat surprisingly, that the single concept that is frequently combined with an active verb in general Swedish usage, lag [law], a clear case of metonymy and a concept with a long tradition of usage with active verbs, is consistently used with the passive in the most central European union legal text.

Concluding remarks
European Union legislation may be expected to influence language usage in all member countries, both by direct references and quotations and by its high prestige (in some cases superseding national law).The constitution in particular is intended for all citizens in all member countries.For these reasons, language expertise in all countries concerned, including Sweden, has devoted efforts to ensuring that each version complies with current linguistic recommendations.
A comparison of the English and Swedish versions on the one hand, and the Swedish version, linguistic recommendations and general Swedish language use, on the other, shows that the Swedish version, by a dominating use of passive verb forms, deviates from both recommendations and current usage.
Legal language is known for certain conservative tendencies which may explain the recurring use of a limited number of possible translation options.In many countries, not least Sweden, lawyers and administrators are working towards more transparent, general-reader friendly forms of language.Practising such principles on European Union texts, in particular central texts aimed at and concerning every single reader alike, should be a natural decision.Following the patterns of English source texts would help translators avoid unwanted sometimes double passives.Grammatical metaphor, particularly in its variety grammatical metonymy, is frequently used in general Swedish language usage and would facilitate the writing, as well as the reading of Swedish European Union texts.

Table 1 .
Number of subject/verb relationships in the English and Swedish versions of the EUC.

Table 2 .
Active verbs in the Swedish version of the EUC, combined with subject.Terms used in the English version in italics after slash, English translations of Swedish terms in square brackets.Only one occurrence when not otherwise marked.To find possible restrictions for the choice of verbs, all active verb forms in the Swedish version of the EUC were listed in Table2.A closer look at the 17 verbs above shows that the majority are stative rather than dynamic verbs, or denote a low degree of activity.Owing to their descriptive character such verbs may be experienced as examples of congruent usage in the above examples, rather than examples of incongruent use, i.e. grammatical metaphor.It is worth noticing that the translator has avoided a more active verb, refer, by writing contains a reference, in spite of the fact that refer, Swedish hänvisa, is often used with abstract subjects like utredningen, [the investigation], paragrafen [the paragraph], rapporten [the report].

Table 3 .
Active verbs in the English version of the EUC, combined with subject.Only one occurrence when not otherwise marked.In comparison toTable 2 it may be of interest to see which verbs are used in the English version, and with what subjects.A total of 47 different active verb forms (types) were registered for 20 different nouns in subject position (types), with law overwhelmingly dominant, 21 types for 116 tokens, to be compared with 21 types of active verb forms for 17 types of nouns in subject position in the Swedish version.These figures are striking evidence of the difference in lexical variability between the two versions.