Article body

1. Introduction

Following Roman Jakobson’s tripartite typology (1959/2012), which recognises intralingual translation (INTRA) as translation on a par with interlingual and intersemiotic translation, and Göpferich (2004; 2007), Schmid (2008; 2012) and Korning Zethsen (2007; 2009) the stance adopted here is that the object field of Translation Studies (TS) should, in a more active way, incorporate INTRA. However, not only has Jakobson’s typology encountered various kinds of criticism, but various approaches to the definition of translation have been proposed which do not square with the typology, usually to the detriment of the translational status of INTRA. The purpose of this article, therefore, will be to review these criticisms as well as other arguments and definitional approaches that would either completely exclude INTRA from the concept of translation, and hence from the object field of TS, or concede it a marginal status only. After discussing and reviewing these points of view, we shall argue for the inclusion of INTRA in a modified version of Toury’s 1995 definition of translation, thereby providing a stipulative, criterial definition of translation to be used for scientific purposes within TS. The article builds on Korning Zethsen (2007) in its motivation and its attempt to argue for the proper inclusion of INTRA, and based on Hill-Madsen (2014), it broadens the discussion and further develops the theoretical arguments. However, before the discussion can be embarked on, it will be necessary to define INTRA. It should be stressed that since our main objective is less to identify what differentiates INTRA from other types of translation than it is to emphasize the commonality of INTRA with other types, the following definition is deliberately kept brief.

If interlingual translation consists in the transcending of a linguistic barrier, i.e. between two different language systems, INTRA must be defined as the crossing of a language-internal barrier. It therefore appears to us that INTRA should – in the first instance, at least – be identified with rewriting between different varieties of the same language, e.g. dialectal (social and/or regional), temporal (i.e. between diachronic varieties) or functional (i.e. between different genres). Examples would thus include the subtitling of geographically peripheral dialects in the standard variety (dialectal INTRA), modern-language versions of pre-modern literature such as Shakespeare or Chaucer (diachronic INTRA), and the rewriting of specialised LSP texts for a lay readership (intergeneric INTRA). Summarising for a new target audience would also belong in this latter category. (For a more elaborate typology of INTRA, see Gottlieb (2008: 56-58)).

2. Arguments against the full inclusion of INTRA in TS

In the following section we will review four main arguments against the full inclusion of INTRA in TS, namely what we have termed the institutional argument, the prototype argument, the equivalence argument, and finally the interlinguality argument.

2.1. The ‘institutional’ argument

A very incisive criticism of Jakobson is voiced by Hermans (1995), who does not dispute Jakobson’s typology as such, but points out that the validity of the typology depends on which of two incompatible perspectives is adopted: an academic or an institutional one. From an academic perspective, Jakobson’s typology may well be acceptable, but if translation is viewed as an institution, the matter becomes very different. The ‘institutional’ status of translation is defined by Hermans as follows: there is such a thing as

a social entity called ‘translation’ and a form of behaviour called ‘translating’ with which, give or take a few nuances, we reckon we are all familiar in our own language and culture. The meaning of ‘translation’ is codified in dictionaries, there are professional activities called translation, we have organizations representing translators, institutes for translator training, etc. It is this ‘public face’ of translation that I have in mind when I speak of translation as ‘institution.’

Hermans 1995: 5

Another aspect of this institutional status is the range of stakeholders involved in translation, such as clients, patrons, translators, agencies, editors and readers, all of whose expectations converge in the norms that regulate the practice of translation (Hermans 1995: 9). One central expectation pertains to equivalence: what consumers of translation expect is a relation of sameness between what they read and the ST (Hermans 1995: 14). It is important to stress that from the perspective of translation as an institution, i.e. from the perspective of ‘the translation consumer,’ the scholarly debate about the (im)possibility of equivalence in translation is irrelevant, because equivalence is generally what external stakeholders expect. The point is that the expectations of stakeholders are constitutive of translation as an institution, i.e. as a ‘social practice’ (Pym 1995: 158) with a specific place and function in society. The institutional expectations are recognised by Pym, who notes that “[…] the translator is an equivalence producer, a professional communicator working for people who pay to believe that, on whatever level is pertinent, A is equivalent to B” (1995: 167), noting also that the expectation of equivalence is “[…] a socially operative belief that enables translations – and translators – to work” (1995: 167) – i.e. enables translation to work as a social practice.

On the other hand, from the point of view of TS, for which translation is not a practice but a research domain, the ontology of the concept may well be different. Thus, as is Herman’s point (1995: 17), in the popular notion translation is only one thing, and that is interlingual,[1] but still there may be good reasons to extend the academic concept of translation to other modes of semiotic derivation, as Jakobson does. This is the stance that will be adopted here, and this is why this article is devoted to arguing for the inclusion of INTRA in the object field of TS. However, as Hermans (1995) points out, the wording of Jakobson’s definition reveals an attempt to span both perspectives, the academic and the institutional one. Following Derrida’s famous deconstructionist criticism of Jakobson, Hermans points to the inequality of status assigned to each of the three types (INTRA, interlingual and intersemiotic): Derrida (1985) lays bare the difference in definitional approach in Jakobson’s essay,[2] noting that INTRA and intersemiotic translation as concepts are each translated intralingually, i.e. reworded in a “definitional interpretation” (Derrida 1985: 173): Jakobson’s own phrasing (1959/2012: 127) was “intralingual translation rewording” and “intersemiotic translation or transmutation.” In the case of interlingual translation, on the other hand, instead of rewording the label, the central word translation is simply repeated (“interlingual translation, or translation proper” (1959/2012: 127)), thus revealing that in spite of the extension of the concept, Jakobson still grants translation proper (TRP) a supreme status: it acquires the meaning of “translation in the ordinary sense,” whereas the other two types are demoted to “translation in the figurative sense” (Derrida 1985: 174). Derrida’s uncovering of this ambivalence in Jakobson’s definitions is what prompts Hermans’ criticism of Jakobson for trying to have it both ways, in a manner of speaking, by offering a typology which reveals the institutional bias towards TRP (‘interlingual translation is what really counts as translation’) while at the same time adopting an academic perspective which extends the perimeters of the concept to other modes (Hermans 1995: 17-18; Pym 2010: 108). No such ‘differential treatment’ of the three translation types will be adopted in this article: it should be emphasised that while full validity must be granted to Hermans’ attribution of institutional status to translation as a social practice and to his recognition that this institution necessarily comes with a rather circumscribed notion of what translation is, the perspective of the present article is necessarily the academic one, with no difference in the status assigned to the three Jakobsonian types.

2.2. The prototype argument

A type of definition which does include INTRA in the concept of translation, but only grants it a marginal status is Halverson’s prototype approach (1999; 2000), which is an interesting contribution to the definitional issue, but in our view nevertheless an untenable one for present purposes. The nature of the approach is the following: Halverson’s point of departure is the observation that previous concepts of ‘translation’ have proven futile in the delimitation of the object field of TS, i.e. have been unable to provide the grounds for a demarcation of ‘translation’ from ‘non-translation’ (Halverson 1999: 2-3), which allegedly pertains to completely relativistic approaches (‘translation can be virtually anything’) as well as to notions of a “completely objective delineation, one “true” delineation” (Halverson 1999: 3). The solution she proposes is to view translation as a prototype concept, i.e. a concept characterised by “membership gradience” (1999: 6), which means that individual subtypes of translation are assigned varying degrees of centrality/peripherality as members of the category.

Philosophically, the notion of graded category membership relies on Wittgenstein’s concept of ‘family resemblances’ from Philosophische Untersuchungen, §§66-67 (1958/1994: 66) (Halverson 1999: 5). This concept is rooted in the observation that attempts to define certain categories in terms of a limited set of clearly identifiable features common to all specimens are futile. Only a complicated network of partially overlapping similarities can be observed (Wittgenstein 1958/1994: 66), and it is for such similarities that Wittgenstein proposes the term ‘family resemblances’ (§67), in the same way as some members of a family (of human blood relations) may share the same hair colour, others the same eye colour, and others the same temperament, etc. (§67), without any single characteristic being common to all, and with some members sharing more points of resemblance than others. The concept of ‘family resemblances’ thus provides Halverson (1999) with the theoretical grounds for positing translation as a prototype category.

Furthermore, Halverson contrasts prototype concepts with the classical (Aristotelian) approach to categorisation. In this approach, categories are believed to match reality independently of the observer, and “the link [between a category and a type of object] is provided by the listing of necessary and sufficient conditions, which match directly with the essential features of the object in question” (Halverson 1999: 5). As a counterargument to the Aristotelian essentialism, Halverson cites modern cognitive research indicating that the content of a given category is in no way conditioned by any inherent properties of objects, but by the cognitive structure of language users: “The evidence put forward seems to point towards one conclusion: categorization is not based on objective qualities inherent in real-world objects, but is dependent on and determined by properties of the human cognizer” (Halverson 1999: 6), which explains why the content ascribed to a given category may vary from one language user to another. Allegedly, it is possible to chart a shared cognitive structure behind a given concept of a language community, but any variation identifiable between the notions of individual users is a manifestation of prototype effects, i.e. different patternings in the way individual ‘cognizers’ attribute varying degrees of centrality/ peripherality to a range of potential members of the category (Halverson 1999: 6).

Several fundamental points of criticism may be raised against the ‘prototype’ solution to the problem of defining translation as the object field of a scientific discipline. First, in referring to Wittgenstein’s concept of ‘family resemblances’ as part of the philosophical grounds for the prototype approach to categorisation, consideration ought to be given to the context in which Wittgenstein’s argument belongs: his argument is aimed at showing how concepts make sense in ordinary or everyday language, without language users agreeing on essentialist definitions of the categories they use. In the everyday life situations in which ordinary language is tied up (Wittgenstein’s so-called language games), we understand each other because we agree on how to use the words. Therefore, even in the absence of essentialist definitions, apparently imprecise concepts (such as game) are completely meaningful in everyday language use (Floor 1982: 185-187; cf. Wittgenstein 1958/1994: 68 (§ 70)). Thus, Wittgenstein’s argument is concerned with the functioning of ordinary language (the well-functioning of it in spite of apparent odds against it), and his aim is not to dispute the possibility of agreeing on more precise conceptual definitions, like those sought in e.g. law and science (Floor 1982: 187). Indeed, Wittgenstein’s argument about family resemblances is hardly relevant to a debate about conceptual definitions within a given scientific discipline. Needless to say, when it comes to scientific inquiry, a minimum requirement is to bring our conceptual tools in order, unequivocal definitions being simply a sine qua non in scientific pursuit (cf. Robinson 2011: 69-70).[3] Therefore, following Robinson, we shall argue that what is needed is a so-called stipulative definition instead (see below), and one that will specify criteria for membership – a definition which might not be in complete agreement with ‘common usage,’ but which may serve as the basis of a scientific taxonomy of translational phenomena.

What Halverson’s inquiry (2000) represents is in effect an attempt to investigate the extent to which a layman’s understanding of translation overlaps with a possible scientific taxonomy of the concept (Jakobson and Toury’s). In contrast, Halliday and Matthiessen (1999) stress the divergence between scientific taxonomies and so-called folk ones belonging to the ‘same’ field of experience, pointing out that the scientific approach differs from the folk one not only in the delicacy of categorisation, but also in terms of classificatory criteria:[4]

The move from folk taxonomies towards scientific ones involves both an increase in steps of delicacy and a change in the criteria used for classification. […The change is one] from overt criteria accessible to the naked eye to covert criteria available only through the application of scientific techniques.

Halliday and Matthiessen 1999: 85-86

Another problematic notion of Halverson’s (1999) is her grounds for rejecting criterial definitions of translation (definitions based on necessary and sufficient conditions). As previously noted, she associates criterial definitions with Aristotelian essentialism, i.e. the conception that such definitions reflect certain inherent properties of the objects that belong to the category. Robinson specifies the Aristotelian notion as follows: “Definition was defined by Aristotle as the essence of a thing […]. On this interpretation, if ‘x is yz’ is a significant and true definition of x, then x is a thing and yz is the essence of that thing” (2011: 154). Robinson, however, dismisses such essentialism, pointing out that “[e]ssence […] is just the human choice of what to mean by a name, misinterpreted as being a metaphysical reality” (2011: 155). In contrast, Robinson maintains that so-called nominal definitions (definitions of the names we give to things) may be valid even without the assumption of any metaphysical reality inherent in the class of phenomena that a given term is chosen to denote. Such definitions are valid simply as a matter of convention, because we may actively decide, especially in science, what content to stipulate for a given term. Definition by stipulation is in fact what Robinson recommends for scientific disciplines: “[w]e must be released from the lexical definitions which merely describe common usage in all its unscientific nature, and allowed to stipulate simpler, more precise, and more unequivocal words and ideas, if we are to have science” (2011: 73). Therefore, a stipulative, criterial definition of translation is what we shall try to provide later.

In 2007, Korning Zethsen argued for an open definition of translation based on prototypology and on the perception of translation as a cluster concept (in line with Tymoczko 2007a). We still do not doubt that prototypology is a valid way of describing the cognitive phenomenon of translation, and the theory provides excellent arguments for a broad perception of the concept of translation with its many facets and family resemblances. In this way an open definition based on prototypology makes sense when the aim is to explain why INTRA is closely related to TRP and consequently why TS may benefit from more research within INTRA and why trained translators in practice would be well-equipped to carry out some of the many INTRA jobs required in modern society. This was indeed the aim of Korning Zethsen (2007). However, in the present article we have another aim: in addition to a more active inclusion of INTRA in TS we wish to argue for, as well as provide, a stipulative definition of translation to be used within TS as a research field. Prototypology may ensure a good understanding of the concept of translation, but will not ensure proper inclusion of INTRA in TS resulting in more INTRA research being carried out.

2.3 The equivalence argument

Although we cannot accept Halverson’s ‘prototype’ solution to the definitional problem within TS, she is right in pointing to the inadequacy of previous criterial approaches in providing answers to the problem of demarcation (the delineation of ‘translation’ from ‘non-translation’). A more detailed scrutiny of one specific type of criterial definition is undertaken by Koller (1995), who demonstrates the limitations of linguistic approaches which equate translational status with equivalence. In Koller’s rendition, these are approaches according to which

translation can be understood as the result of a text-processing activity, by means of which a source-language text is transposed into a target-language text. Between the resultant text in L2 […] and the source text in L1 […] there exists a relationship, which can be designated as a translational, or equivalence relation [emphasis as in the original].

1995: 196

Though favourable to linguistic approaches himself, Koller demonstrates the untenability of this position by pointing to what he terms the “double linkage” of translations: on the one hand, the link of the TT with the ST whose content the translation is supposed to represent, and on the other hand the link with the target language audience (1995: 197). This, of course, is identical to the dual, and often irreconcilable, set of obligations incumbent on the translator: the practice of translation as a perennial matter of divided loyalties between fidelity to the ST and consideration for the TL readership (cf. Tymoczko 1999: 55). Koller points out what is well-known to any practising translator, namely that equivalence in the form of faithfulness to the source text is in many cases inadequate if a TT is to succeed in providing its readers with proper access to the content of the ST: “If translations are to be understandable, or rather, if they are to convey certain values of the source-language text to the target-language reader, this cannot but entail the application of text-revision methods” (Koller 1995: 205). In other words, a well-functioning TT is often the product of translational procedures that depart from strict adherence to ST wordings, by modifying and sometimes expanding the semantic content of the original. The point, however, is that such “text-revisional elements,” as Koller calls them (1995: 208), are not “equivalence-guided text reproduction” (1995: 204), but free text production,[5] i.e. ‘free’ elements only indirectly related to the ST, which means that theories of translation which hypostasize equivalence as a demarcation criterion are bound to exclude significant elements of texts that we normally regard as translations (cf. Schmid 2008: 31; cf. Toury 1986: 1120).

The view of equivalence as something unattainable is also implicit in Tymoczko (1999). Like Koller, she stresses the dual orientation of any translation (towards the ST and the TL readership at the same time (Tymoczko 1999: 56)), but she goes one step further by pointing to a specific trait of all translation which she metaphorizes as ‘metonymic’ in reference to the rhetorical device of metonymy (where e.g. a part of something is made to stand for the whole). This is what captures the realities of translating: translations can never be anything but partial, i.e. can only ‘stand for’ part of the original content, since translating always – and no matter what type of texts are involved – entails selecting some aspects of the ST to represent in the TT and ignoring others (1999: 55; see also Tymoczko 2007a: 36-37). Another interesting point of Tymoczko’s which deserves mentioning is her observation that when it comes to familiar dichotomies such as Toury’s adequacy vs. acceptability (e.g. 1995: 57) and Venuti’s foreignization vs. domestication (1995), “there is no single polarity that describes the orientation of a translation, no single positioning along a linear continuum” (Tymoczko 1999: 56): parts of a translation may closely adhere to the ST while others may be much more oriented towards making the TT ‘digestible’ to the TL readership (1999: 55). However, although a translation may undeniably be more or less oriented towards rendering ST content faithfully and conversely less or more oriented towards TL accessibility, there is a deeper sense in which translational equivalence is in the final instance impossible, and in which translation is always in the last resort oriented towards the target audience, even in spite of an intended ST orientation. In Saussurian terms, replacing the signifiants of a ST with those from another language does not mean that the original signifiés come along as fellow travellers on the journey across the linguistic divide, arriving at their destination to embody themselves in the signifiants of the TT. In the words of Derrida, ‘the signified is inseparable from the signifier’ (1972/1981: 18), which means that nothing in fact moves (cf. Chesterman (1997: 8), from whom the travelling metaphor has been borrowed. Cf. also Stecconi (2007: 21)). What happens in translation is that one text sparks another text into being (Chesterman 1997: 8), and once TL signifiants have been selected, they signify TL signifiés, embedded as these are in the TL cultural framework (cf. Snell-Hornby (1998/1995: 42). Obviously, ST signifiés are (in most cases, at least) what prompt TL signifiants to be chosen, but once chosen, the TL signifiants signify TL meanings which do not per se ‘refer back to’ ST meanings; only by social agreement, such as through the legal institution of notarization, is it possible to claim that they do. We therefore fully concur with Derrida, in whose view “we will never have, and in fact have never had, to do with some “transport” of pure signifieds from one language to another, or within the same language, that the signifying instrument would leave virgin and untouched” (1972/1981: 20). Derrida thus rejects the idea of what he terms “the transcendental signified,” drawing the conclusion that translation equals transformation (1972/1981: 20). Our position is thus what Chesterman and Arrojo (2000: 151) term non-essentialism, i.e. the view that meanings are always context-bound and that there are no stable, objective meanings that let themselves be ‘carried across’ from ST to TT (for a similar view, see also Hermans 1996). It is in this sense that translation is hermeneutics, since what happens in translation is that TT meanings make sense of ST meanings through replacement (cf. Steiner’s (1975/2012) conception of understanding as an act of translation).

The culture-bound nature of meaning is nothing new, of course, and the inherent problems of translation have long been recognized as a fundamental obstacle in e.g. ethnography and cultural anthropology (cf. Hermans 1996: 16; cf. Rubel and Rosman 2003). Within TS, Snell-Hornby (1988/1995: 41) has pointed out the corollary of the non-essentialist position, namely that translation becomes logically impossible, at least if semantic identity is stipulated as criterial to translation. However unpleasant the non-essentialist position is to translation as an institution (in Hermans’ sense, as previously outlined), we nevertheless believe it is inescapable for TS as a scholarly discipline. What the position hardly precludes, however, is the ability of translation to facilitate communication and at least a satisfactory degree of understanding. In spite of the culture-bound nature of meaning, it would appear that translation still affords the possibility of adequate semantic approximation which may be sufficient for communication to succeed (cf. Kaiser-Cooke 2004: 195). In this way, the issue of ST- versus TT-orientation becomes an issue of approximation degree. It must be conceded that most modern translation scholars in fact appear to acknowledge the impossibility of equivalence in the sense of ‘complete semantic identity’ (cf. Chesterman 1998: 27). In the words of Juliane House,

[v]iews of equivalence as simply based on formal, syntactic, and lexical similarities have been criticised for a long time – not least because it has long been recognised that any two linguistic items in two different languages are multiply ambiguous. Further, purely formal definitions of equivalence have long been revealed as deficient in that they cannot explain appropriate use in communicative performance. This is why functional, pragmatic equivalence has been a concept accepted in contrastive linguistics for a long time, and it is this type of equivalence which is most relevant for translation.

House 2001: 135-136

Yet, because ‘equivalence’ implies ‘identity’ no matter how the concept is modified, there is still no guarantee that equivalence even in this expanded sense (as pragmatic/functional) is possible. The assumption that two different texts (a ST and a TT) can have the ‘same’ effect on two different audiences is in fact what Chesterman denounces as “the homogeneous readership fallacy” (1997: 35). In our view, a better solution is to dispense with the concept of equivalence altogether as a definition criterion.[6]

Nevertheless, the importance of some kind of ST-TT correspondence is undeniable if a given semiotic entity is to count as a translation of another, anterior entity. This applies to any kind of semiotic derivation, and the argument is common-sensical: for a derivational relationship to obtain between two texts (in the broadest sense), some kind of similarity must logically exist (cf. Stecconi 2004: 479). Similarity, and not equivalence, is in fact what several translation scholars like Chesterman (1996; 2007b; 2007a), Yallop (2001), Stecconi (2007: 21), Tymoczko (2007a) and Korning Zethsen (2007) emphasize as relevant in translation. Thus, in the words of Tymoczko (2007a: 32), “[…] equivalence in translation theory and practice can only be a useful concept when it is understood as a form of similarity.” When it comes to any closer designation of the nature of similarity in translation, Tymoczko (2007a) is rather vague, whereas Chesterman (1996; 1998: 12-16; 2007a: 61-62) provides an incisive analysis of the concept of similarity as such, showing that the concept is not a uniform one, but that a distinction has to be made between two types: convergent and divergent similarity. Of these two types, convergent similarity is bi-directional: it means that two entities can be deemed to be similar from the perspective of either: A in this case is similar to B as much as B is similar to A. In Chesterman’s notation: A ↔ B (1996: 161). In the case of divergent similarity, the resemblance is unidirectional: it means that if one entity is derived from another, the product of the derivation can be recognised as being similar to the original, but the converse is not necessarily true, i.e. the original may not be judged to be similar to the derived product. In fact, an entity may function as an original from which many secondary entities may be derived, all of which bear some resemblance to A, but which are nevertheless different from each other. In Chesterman’s notation: A → A’, A’’, A’’’ (1996: 161). The relevance of this analysis to TS is obvious: similarity in translation is almost always divergent, “asymmetric” (Pym 2010: 26) or “irreversible” (Toury 1986: 1117), and one ST may give rise to a number of different types of TTs, all of which will be similar to the original in some way.[7] This observation is perhaps especially relevant to INTRA as the derivational relationship between ST and TT is often not as close as in prototypical interlingual translation, meaning that a TT is recognizably derived from its ST, but any kind of ‘back translation,’ i.e. an attempt to reconstruct the ST from the TT would be virtually impossible.

Moreover, in accordance with skopos theory, Chesterman suggests that what matters in translation is relevant similarity, and that what counts as such in a given translation assignment is dependent on the prospective role of the TT in the communicative situation in which it is to function (2007b: 68).[8] In Chesterman’s own words,

[a]translator submitting a translation to a client claims, in effect, that the translation is similar to the source text in a way that is relevant to the client’s point of view. Without a point of view, there is no way to assess the relevance of the similarity; indeed there is no way to perceive any similarity in the first place. There is always a point of view, even though it may not be explicitly stated or acknowledged.

Chesterman 2007b: 68[9]

2.4 The interlinguality argument

Apart from the ‘equivalence’ argument critiqued above, the extension of the concept of translation to include INTRA runs counter to a conception that would confine translation to text operations involving two different languages. Newmark, for example, adamantly excludes INTRA from translational status because “[…] the qualitative difference between ‘interlingual’ and ‘intralingual’ translation is so great that it makes a nonsense of the concept of translation” (1991: 69). Likewise, Mossop (1998: 252) is at pains to exclude INTRA from translation, and Schubert (2005: 126) asserts that “[t]o translate means to render a text into a different language. Translation is by definition interlingual.”[10] Even in Toury’s – otherwise very broad – definition of translation, interlinguality is in fact close to being criterial: “Regarding a text as a translation entails the obvious assumption that there is another text in another culture/ language which has both chronological and logical priority over it” [our emphasis] (1995: 33-34). However, the qualification of language as “culture/language” may be taken to indicate that interlinguality may be too restrictive a criterion.

In a cogently argued refutation of notions such as those cited above, Schmid (2008) points out that the fuzziness of languages as separate entities makes interlinguality a far from solid basis on which to build a demarcation criterion of translation. He thus points out that the boundaries between languages in many cases reflect socio-political conventions, and not any linguistic characteristics. Thus, although Serbian and Croatian are to all intents and purposes one and the same language, or two dialects at the most, political realities in former Yugoslavia have given rise to a demand in the two populations that the two dialects should be treated as two separate languages (Schmid 2008: 60; cf. Wardaugh 1986: 33). Such politically determined conventions impose an artificial barrier that defies the criterion of mutual intelligibility as the means of distinguishing languages from dialects (two linguistic systems are to be regarded as dialects if they are mutually intelligible and as languages if they are not (Matthews 2005: 96-97)). In the same vein, Schmid (2008), Matthews (2005: 97) and Wardaugh (1986: 28) all mention the example of Danish and Norwegian, which are by convention recognized as two different languages, but whose speakers are in fact able to understand each other, at least with a little effort. A converse example mentioned by Wardaugh (1986: 28) is that of Chinese dialects, such as Mandarin and Cantonese, which are in fact mutually unintelligible, but nevertheless regarded as dialects only. Yet another adducible phenomenon which blurs the division between languages and dialects is the fact that two dialects on each side of a national border, recognized by convention as belonging to different languages, may well have more characteristics in common than two geographically distant dialects within the ‘same’ language. Thus, Germans living close to the Netherlands may find that their dialect has more in common with Dutch than with southern variants of German (Wardaugh 1986: 28).

Altogether, the fact that languages shade into dialects and vice-versa is a convincing argument against any attempt to uphold interlinguality as a demarcation criterion for the concept of translation, and it provides strong grounds for including especially INTRA between regional/social dialects in the concept of translation. Whether the language-dialect continuum provides grounds for extending the concept of translation to include INTRA between functional dialects, i.e. genres, is a different matter. For this purpose, an emphasis on translation as a cross-cultural transfer (or rather replacement) operation and not simple transcoding between linguistic systems may be a better avenue. Schmid (2008), in fact, extends the ‘cross-culturalness’ of translation to INTRA as well, pointing out that cultural borderlines are not restricted to those running between cultures encoded in the semantics of a ‘national’ language. In the same way as Halliday (1978: 99; 123) views cultures as systems of meaning, Schmid (2008: 44) conceives of culture as a conceptual system, but at the same time points out that the system of meaning constituted by a given ‘national’ language is far from a uniform entity – that even within a language, there are competing conceptual systems, i.e. conflicting semantic construals of the world of experience: “There are a lot of smaller cultures within a “language community” that conceptualize aspects of the world differently and thus have to recur to processes of translation in order to guarantee successful communication among each other” (Schmid 2008: 48).[11] A good example of a ‘subculture’ in this sense is of course a scientific community (cf. Aikenhead 2001: 24-25). A person typically belongs to a series of different cultures (or discourse communities as Swales (1990) would call them): “[s]ince each person takes part in a variety of […] domains of interaction and experience, a number of concept systems rooted in all kinds of different settings converge in a person’s cultural make-up” (Schmid 2008: 47). As previously indicated, Schmid locates the need for translation in the divergence between semantic systems, which he regards as the principal type of communication barrier that necessitates translation: “The need for translation arises when people do not share a critical mass of mutual accessibility in their conceptualizations and consequently cannot refer to a common conceptual framework in which to make sense of each other’s utterances” (Schmid 2008: 51). In Schmid’s understanding, therefore, translation “comprises the reconfiguration of concepts from the perspective of another concept system” (Schmid 2008: 54),[12] which echoes our previous analysis of translation as the semiotic act of capturing one set of concepts in terms of an alternative set. As a corrective to Schmid’s generalization, it may be necessary to point to the truth in Mossop’s observation (1998: 252) that the need for translation may in the first instance arise from the foreignness of the lexicogrammar of a ST, but Schmid is quite right in extending communication barriers requiring translation to those which derive from incompatible conceptual systems, whether or not the lexicogrammar of the two systems are those of the ‘same’ (‘national’) language.

3. A criterial definition of translation – discussion and conclusion

A possible shortcut to the inclusion of INTRA in the object field of TS would be to join ranks with those who represent completely relativistic approaches to the definition of translation. The most famous exponent of this view is probably G. Toury, according to whom TS should preoccupy itself with “all utterances which are presented or regarded as such [i.e. as translations] within the target culture, on no matter what grounds” (1995: 32), which mirrors the conception that “there is no pretense that the nature of translation is given, or fixed in any way” (1995: 32). In other words, translation is defined by the assumptions of any given target culture. Another similar approach is that of Tymoczko (2007b), who lauds Toury’s approach for releasing the concept of translation from what she denounces as Eurocentric models (2007b: 81), and supplements his relativistic stance by asserting that “there are no necessary and sufficient conditions that can identify all translations and that at the same time exclude all non-translations across time and space” (2007b: 78).

Whether we lay ourselves open to accusations of Eurocentrism or not, we hold that it is in the best interest of TS as a research field to reject relativistic stances like those above, the reason being that any field of research wishing to be taken seriously as an academic discipline must satisfy the minimum requirement of being able to delimit the object of study, and this is only achievable by proposing a criterial definition.

Moreover, such a definition need not necessarily be at odds with an aim of broadening the object of field of TS beyond narrow, traditional European conceptions of translation, which appears to be a primary concern of Tymoczko’s. Apparently, she fears that upholding a criterial definition will automatically result in the exclusion of rewriting types that non-European cultures would accept as translation (1999: 60-68). However, the reverse may well turn out to be the case, since in relying on the concept of ‘assumed translation,’ Toury’s stance invites folk notions to interfere with a scholarly definition, and there is no guarantee that such folk notions (European or non-European) may not be more exclusive than a scholarly approach, contrary to the intentions behind the relativistic stance. INTRA, for instance, would probably be almost totally excluded in the present Danish culture, and the TS research community could lose valuable insights. The example is of course taken from a European context, but who says that a similar rejection of the translational status of INTRA could not be encountered outside Europe?

What is more, the relativistic stance adopted by Toury is in effect defeated by his own introduction of the three ‘postulates,’ which in effect represent Toury’s own intervention, so to speak, in this culturally relative process of ‘assuming.’ The postulates must be read as Toury’s own specification of the grounds for ‘assuming,’ i.e. for regarding a given text as a translation. He thus ends up contradicting his own pretense at relativism by providing what is actually a criterial definition (cf. Pym 2010: 85). Following Chesterman (1997: 62) and Korning Zethsen (2007), we take the view that Toury’s three criteria, or “postulates,” as he terms them, are in fact those which remain valid (subject to certain modifications, to be proposed below). The three criteria are 1) the Source Text Postulate 2) the Transfer Postulate and 3) the Relationship Postulate (Toury 1995: 33), each being defined as follows:

  1. The Source Text Postulate

    Regarding a text as a translation entails the obvious assumption that there is another text, in another culture/language, which has both chronological and logical priority over it: not only has such an assumed text assumedly preceded the one taken to be its translation in time, but it is also assumed to have served as a departure point and basis for the latter. […] (Toury 1995: 33-34)

  2. The Transfer Postulate

    The Source-Text Postulate also entails the assumption that the process whereby the assumed translation came into being involved the transference of certain features from the assumed source text which the two texts now share. […] (Toury 1995: 34)

  3. The Relationship Postulate

    Finally, adopting the assumption that a text is a translation also implies that there are accountable relationships which tie it to its original, an obvious function of that which the two texts allegedly share and which is taken to have been transferred across the cultural-semiotic (and linguistic) border. […] (Toury 1995: 34-35)

The second and the third of these postulates cannot be accepted outright, however, and the objection concerns Toury’s insistence on transfer (or “transference,” as he terms it). As previously shown, the idea of something – entities of meaning – moving from ST to TT is untenable, because translation is not a matter of ‘moving’ content, but a matter of producing a new text on the basis of an anterior one. We therefore propose the following modification of the two postulates: instead of transfer, derivation should be posited as a criterion,[13] and, following Chesterman, relationship should be specified as one consisting in relevant similarity (cf. Chesterman 1997: 62), brought about through the application of derivational strategies. As previously argued (in accordance with Chesterman 2007b), what counts as relevant similarity in a given ST-TT pair is skopos-dependent. Moreover, the criterion of relationship must be further expanded to encompass two additional aspects pointed out by Stecconi (2004: 479-482; 2007: 23-24),[14] namely mediation and semiotic difference, both of which are to be conceived of as sine qua non to translation (2004: 483). Thus, translational status logically entails mediation, that is a mediating function on the part of the TT in relation to ST semiotic content. In Stecconi’s own words, “there is no translation if the target sign does not speak on behalf of the source sign” (2004: 482). Moreover, Stecconi points to ST-TT semiotic difference as an inherent feature of translation, i.e. that translation logically consists in the transcending of a semiotic barrier, or a “fold between semiotic systems” (2004: 480).

What remains to be considered is whether INTRA in its different forms is compatible with these modified Touryan criteria, to which the answer must be in the affirmative, except in one respect, as Korning Zethsen (2007: 293) points out, since linguality is in fact part of the first of the three postulates. This becomes especially clear in Toury’s own gloss, according to which a ST is required to belong to “another culture and language [our emphasis]” (1995: 35) as compared with that of the TT. Nevertheless, contra Toury, it must be maintained that linguality cannot be elevated to criterial status, as previously argued. Culture, on the other hand, has been shown to be intimately linked with communication barriers necessitating translation, but it has also been argued that since cultural differences derive from divergent conceptual systems, the conception of cultures as co-extensive with ‘national’ languages only is much too simplistic. Moreover, as indicated above, the view taken here is that the concept of translation should be expanded, but not confined to semiotic transformation between cultures, even in this extended sense – should be expanded, in fact, to include semiotic conversion across communication barriers at any or all strata of language (in the case where ST and TT alike consist in verbal language): conceptual (i.e. semantic), lexicogrammatical and phonological/graphological.

We therefore maintain (a slightly adjusted version of) the definition proposed in Korning Zethsen (2007: 299) for the delimitation of the scientific field of TS. A translation is a text which conforms to the following conditions:

  • A source text (verbal or non-verbal) exists or has existed at some point in time.

  • The target text has been derived from the source text (resulting in a new product in another language, genre, medium or semiotic system).

  • The resulting relationship is one of relevant similarity, which may take many forms depending on the skopos.

Obviously, the definition is pillared on certain key concepts, such as ‘source text,’ ‘derivation’ and ‘relationship,’ which may themselves require a second round of definition. This, however, is beyond the scope of the present article, but we do believe that these concepts, too, can and should, in principle, be given a criterial definition. On the other hand, there is no denying that the application of these concepts to real-life phenomena is likely to be fraught with difficulties, and that real-life instances are almost certain to form a cline from the prototypical to the peripheral. It could therefore be said that the definition relies on open concepts (in Wittgenstein’s sense), but not on folk assumptions. The concepts are “open” (which is why Korning Zethsen (2007: 297) terms it “an open definition”) in the sense that they may be more or less prototypical in relation to a certain culture, at a certain time and in a certain context, but the definition itself is a criterial one which may be helpful in delimiting a discipline like TS. A broad and inclusive definition, but not a definition where translation can be virtually anything. Important to the purposes of this article is the fact that the definition encompasses not only INTRA in its many variations but also highly heterofunctional TRP, i.e. a translation where the skopos demands a much-changed version, localisation, précis-writing, expert-to-layman communication, etc.) which would seriously challenge or in most cases exceed the boundaries of what theory traditionally perceives as TRP. As Steiner points out (1975: 260-61) INTRA and TRP raise issues of the same order and are, at crucial points, similar. This means that TS may well lose out on valuable insights if INTRA is in effect excluded from the discipline. In the practical and especially the didactic world, a too narrow definition of the field only sets an artificial boundary for translators and the jobs they see themselves as able to carry out. In today’s world a quite narrow version of TRP is considered prototypical translation, both by people in general and by the field of TS, but we hope that translation scholars will increasingly include INTRA in their research and that the present article will contribute to such a development.