Features of Discourse Presentation in Translation : Literary and Narratological Insights into Translation Universals

Several translation scholars have recognised translation as a form of discourse mediation or discourse presentation (see, for example, Mossop 1998). In line with this, ‘universals’ of translation have also been re-framed in the larger context of discourse mediation, as mediation universals rather than something strictly translationspecific (Ulrych 2009). In the present article, this line of enquiry is developed by comparing some of the alleged universals of translation, namely standardization and explicitation, with insights from literary and narratological studies on the nature of discourse presentation. The notion of reportive or interpretative interference (Sternberg 1982) and Fludernik’s (1993) claim that all represented discourse is typical and schematic in nature seem to bear curious resemblance to the notion of standardization or normalization, posited as a possible universal of translation (Mauranen & Kujamäki 2004). Drawing on the results of my earlier research (Kuusi 2011), I present examples of free indirect discourse (FID) used in Dostoevsky’s novel Crime and Punishment with their translations into Finnish. Analyzing the translations, I demonstrate how in translations, the narratological and literary-theoretical notions of reportive interference and typification/schematization coincide with the translation-theoretical notions of explicitation and standardization.


Introduction
In the present article 1 , discourse presentation is approached from an interdisciplinary perspective, combining insights from translation studies, narratology and literary theory.
Several translation scholars have recognised the nature of translation as discourse mediation or discourse presentation (see, for example, Mossop 1998, Ulrych 2009and Taivalkoski-Shilov 2010).The term discourse mediation attains to a text as a whole, while discourse presentation, also referred to as speech reporting, speech/discourse representation and reported speech, focuses on the segmental level. 2 Both discourse mediation and discourse presentation present As early as the 1930s, Vološinov (1972Vološinov ( [1930]]) posited the inevitable impact of the mediating agent on the mediated discourse.This impact, later coined reportive interference by Sternberg (1982), together with the communicative pressure present in discourse mediation, renders a simple reproduction practically impossible and leads to modifications or adjustments of the original discourse.In post-classical narratology, Fludernik (1993) claims that all represented discourse is typical and schematic in nature.Instead of a mimetic reproduction of the original, represented discourse constitutes a typified representation of the original, a version that better suits the new communicative context.This typifying tendency seems curiously similar to tendencies proposed in translation studies as possible universals of translation, such as normalization, conventionalization, simplification and standardization (see, for example, Mauranen & Kujamäki 2004).Drawing parallels between translation universals and the nature of discourse presentation allows us to see translation universals more like regularities of discourse mediation/presentation rather than something strictly translation-specific.
In this article, discourse mediation is approached on two levels: on the intralingual level, as speech representation in literary narratives, and on the interlingual level, as translation.In translations of discourse presentation in literary narratives, these two coincide with each other.In the following sections, I examine literary translations of free indirect discourse (FID), a mode of discourse presentation widely used in literary fiction to merge the perspective of the reporting speaker (the narrator) with that of the reported speaker (the character).In FID, indices of indirect discourse (ID), such as third-person reference to the subject of consciousness, typically coincide with features of direct discourse (DD), such as exclamations or direct questions.Hence, FID foregrounds the duality present in all discourse presentation, consisting of the presence of two voices and points of view in the same utterancethe reporting and the reported.
In Kuusi (2011), I compared passages containing FID from Dostoevsky's novel Prestuplenie i nakazanie (Crime and Punishment) with six Finnish and two English translations of the novel.In these translations, FID was often substituted with ID or DD, or simply rendered less discernible than in the original.Furthermore, my analysis also revealed that all of the linguistic changes resulting in the weakening or loss of FID in translations either normalized the text, or 'explicitized' it, i.e. rendered it more explicit.In the present article, I present an analysis of two passages of FID from this data and their translations into Finnish.By analyzing these translations, I demonstrate how the narratological and literary-theoretical notions of reportive interference and typification/schematization coincide with the notions of explicitation and normalization.

Discourse mediation and discourse presentation
Translation has been conceptualized as both a form of mediated discourse, along with other forms of discourse mediation such as summarizing and editing (Ulrych 2009; see also Lefevere 1992), and as a mode of discourse presentation (otherwise called reported speech), in parallel with, for example, direct and indirect discourse (Mossop 1998; see also Hermans 2007: 65-76).Whereas the term discourse mediation is typically used on the level of whole texts being mediated for new audiences and new contexts, speech/discourse presentation is employed on the level of single utterances embedded in other utterances.The difference between the two, therefore, is one of proportion between the anterior and the posterior discourse.In discourse presentation, the prior discourse typically constitutes only a part of the new discourse, while in discourse mediation the mediated text as a whole is an adjusted version of the previous text.Both, however, refer to representations 3 of a pre-existing discourse.In Ulrych (2009: 222), forms of discourse mediation are described as "recontextualizations of pre-existing discourses", whereas Short et al. (2002: 334) describe discourse presentation as "report of anterior speech".The two may be treated as largely overlapping, if not coinciding.In the present article, they are approached as different aspects of the same phenomenon, with the broader term discourse mediation concerning the text as a whole, while discourse presentation focuses on the segmental level.Vološinov's (1973) classical definition of reported speech is "speech within speech, utterance within utterance, and at the same time also speech about speech, utterance about utterance" (Vološinov 1973: 115;in Russian Vološinov (1972[1930]: 113).The first part of his definition illustrates the nesting structure of discourse presentation, bringing to mind the image of a matryoshka doll, where one object contains another similar object.The second part of Vološinov's definition draws attention to the reporter's imprint on the reported discoursethe fact that the reporting speaker is not merely mediating, but also 3 Short et al. (2002: 336) insist on the importance of maintaining a clear distinction between the terms ‛report', 'presentation' and 'representation' in the context of discourse presentation.In their view, 'report' and 'representation' imply a pre-existing discourse, whereas 'presentation' (predominantly used in the stylistic analysis of literature) does not.In the present article, however, the notions are applied to linguistic mediation both with and without an existing anterior discourse (translation, fictional discourse).The term discourse presentation is preferred, but the choice of term in each case reflects the terminology employed in the referenced research.
has something to say about the discourse being presented or mediated.This imprint, whether explicit or implicit, intended or unintended, is an integral part of discourse presentation, for, as Vološinov puts it, discourse presentation inevitably entails "active reception of other speakers' speech" (Vološinov 1973: 116-117;in Russian Vološinov (1972[1930]: 114-115).
In the present article, two categories of discourse presentation receive special attention: direct discourse, in line with Mossop's (1998) definition of translation as free direct quoting, and free indirect discourse, a mode of discourse presentation praised for its ability to accommodate the point of view of the reporter and the reportee in the same utterance (for a detailed account of this dual-voice quality of FID, see Pascal 1977).

Translation as discourse mediation/presentation
To describe translation simply as communication does not capture its complexity.Indeed, the nature of translation as mediated or represented discourse has often been recognized by various translation scholars.In translation, we are dealing with communicating a pre-existing discourse to a new audience, and therefore with a form of constrained communicationmore specifically, communication constrained by the variables of the communicative situation such as the source text, the mediator, the assigned function and the addressee of the translation.Mossop (1998) proposes a definition of translation production as free direct quoting, his aim being to define translation in opposition to other forms of communication and language mediation, and his definition includes the processing of the wording of the source text in sequential chunks with an imitative purpose (Mossop 1998: 231, 251-253, 261).For the purpose of the present article, Mossop's definition is most informative.By quoting, Mossop does not refer to a verbatim report, but to a selective demonstration (not description) of the original discourse (ibid. 244-245).The imitative purpose presupposes a pre-existing source text and sets translation apart from hypothetical speech representation, whereas the sequential processing of the wording separates translation from the thematic re-processing of a text as a whole, as is the case, for example, in oral narration (ibid. 247-248, 251).The definition of translation as free direct quoting implies that translation preserves the deictic orientation of the original, such as the first person forms referring to the original speaker.The freeness, on the other hand, presupposes that there need not be a pre-or post-quotational phrase such as she says accompanying the translation (ibid.244, 251).
Mossop's definition is a theoretical one with no practical prescription implied.Therefore, it presupposes the possibility of categorizing acts of translation as centrally or marginally translational, depending on how well they fit the criteria set out above (Mossop 1998: 250, 262).Thus, for example, when a translation in a real communication situation describes a speech act instead of demonstrating it, Mossop considers it marginally translational, with no value judgement implied (ibid. 245-246).
Taivalkoski-Shilov (2010), however, sees the direct-indirect distinction, and the consideration of indirect translation as marginally translational, as a shortcoming in Mossop's definition.She argues that this is the case based on actual translations sometimes in fact employing indirect reporting rather than direct quotation (Taivalkoski-Shilov 2010: 2).I shall return to the question of directness/indirectness in section 6.
In Ulrych (2009), translation is discussed as a form of discourse mediation, along with editing and non-native language production.In Ulrych's approach, all forms of discourse mediation are seen as recontextualizations, with a preexisting text being processed to meet the needs of a new audience, and therefore essentially recipient-oriented (ibid. 222, 224).This orientation leads to a selective rather than a verbatim mediating manner, picturesquely described by Ulrych as refraction (Ulrych 2009: 224).
The ultimate aim of Ulrych's analysis is to determine whether the other forms of discourse mediation share the same modifications or adjustments which are characteristic of translation, and conventionally referred to as translation universals (Ulrych 2009: 222-223).While Mossop seeks to define translation as a distinctive form of reported speech, Ulrych's aim is to highlight the similarities between translation and other forms of discourse mediation.The two approaches have opposite goals; combined, they offer a more comprehensive view of translation as discourse presentation/mediation.

Universals of translation as universals of discourse mediation
Approaching translation as an act of discourse mediation opens up new avenues for the study of the regularities of translational behaviour, referred to as translation universals (see, for example, Baker 1993 and1996;Mauranen & Kujamäki 2004).It is assumed that these regularities characterize translation in general, irrespective of the language, text type or situational factors involved.Explicitation, normalization, simplification and standardization are amongst the most examined features of translation promoted as possible translation universal candidates.In other words, translations are often construed as being more explicit, normal (conventional) and simpler than non-translated texts, such as their source-texts or texts originally written in the target language (see, for example, Laviosa 2009: 306-309;Shuttleworth & Cowie 1997: 193-194).
Although stimulating research has been carried out on translation universals, their status as genuine universals of translation remains inconclusive and still requires further empirical evidence (see, for example, Mauranen & Kujamäki 2004).Thus, in translation studies, the suggested universals of translation are much-debated hypotheses rather than accepted facts.Furthermore, the terminology in the field is problematic and in part overlapping.In fact, even the very term universal is open to debate.It can be understood either in an absolute sense, or in a probabilistic or typical sense (see, for example, Toury 2004: 17-21;Malmkjaer 2011: 87).In the present study, the latter interpretation is adopted.Olohan (2004: 92) points out that until we know for sure that we are dealing with genuine universals instead of norm-based features, we should refrain from using the term universal and refer to them simply as features of translation.Other suggested terms for these features include laws (Toury 2004: 28-29) and regularities (Chesterman 2008: 367), these different terms reflecting researchers' various ontological stances.This wealth of alternative terms reflects the status of research as a work in progress, and a consensus on terms is likely to be reached only after we have more decisive evidence on the nature of the phenomenon.However premature, in the present paper I will use the term translation universal as the most established of the terms and hopefully one that most readers will recognize.
In addition, Chesterman (2004: 44) notes further terminological difficulties, since terms such as standardization, normalization and simplification all seem to denote the same phenomenon.Pym (2008) actually suggests that both simplification and normalization fall under the umbrella of the higher-level concept of standardization (on standardization, see Toury 1995: 267-274).Often the actual textual manifestations of these tendencies seem to overlap.For example, normalization of punctuation has been taken to indicate normalization in some studies and simplification in others (Pym 2008: 318).Similarly, lower lexical density and lower type-token ratio, both regarded as indications of lexical simplification, could also be considered indications of explicitation, which adds grammatical words (leading to lower lexical density) and common words (lower type-token ratio) (ibid.318; see also Halverson 2003: 221).Likewise, in most studies, a preference for high-frequency words in translation has been taken as an indication of lexical simplification.Halverson (2003: 236), however, notes that this also conventionalizes the text, for conventionality includes a preference for higher-frequency items.To maintain normalization as a separate category seems especially challenging, for its manifestations must necessarily vary, depending on what is considered normal or conventional in the target language.For example, in Becher's study on explicitation (2010), the target language (German) conventions require a higher level of explicitness than those of the source language (English).This leads Becher to think that a number of the instances of explicitation are caused by the German stylistic conventions concerning explicitness and, therefore, cannot qualify as instances of universal (translation-inherent) explicitation (Becher 2010: 8-19).However, due to the different conventions concerning explicitness in the source and the target languages, such instances could just as well be taken to indicate normalization (a tendency not discussed by Becher) (Kuusi 2011: 146).
In addition to overlaps in their manifestations, the tendencies of explicitation, normalization, simplification and standardization seem to share the same goal.Standardization and normalization (the latter alternatively referred to as conventionalization) are clearly synonyms, and both are described as tendencies towards conventionality and readability, aimed at lessening the reader's burden in processing the text (Baker 1996: 183-184;Laviosa-Braithwaite 1998: 290).Similarly, explicitation and simplificationin spite of being treated as separate categoriesare both described by scholars as tendencies that increase the readability of the text and thus help the reader to process it (Pym 2008: 318; see also Saldanha 2008: 32;Leuven-Zwart 1990: 81;Baker 1995: 236-238 and1996: 183).This shared goal lends support to the claim that the tendencies should be treated not as separate phenomena, but rather as different manifestations of the same general tendency to standardize (or normalize, or conventionalize, whichever term one prefers).
If, however, any of the suggested translation universals were in fact proved to be typical of translation in general, such findings would still beg elaboration.Both cognitive and pragmatic explanations have been proposed for translation universals.Chesterman (2004: 44) argues that in the end, the causes are to be found in the cognitive processing of the translator.After all, the translator's interpretation is the filter through which all the information is (or is not) transferred to the translation.Halverson (2003) delves deeper into the cognitive domain, explaining the universals of simplification and normalization through cognitive constraints in human information processing.In her analysis, based on Langacker's cognitive grammar, the choice of lexical items in translation is influenced by cognitively salient features such as category prototypes, (features occurring with high frequency) (Halverson 2003: 218-219, 221).Pym (2008), on the other hand, explains translation universalsnot only standardization and its sub-categories, but also the opposed tendency of interferenceas resulting from the translator's pragmatic need to avoid communicative risks.Opting for the most normal, and in that sense safest, choice leads to standardization, while the opposed strategy of interference shifts the responsibility from the translator to the source text (Pym 2008: 324).
Although not formulated in terms of discourse mediation, both these explanationsthe mechanisms of the cognitive processing of information, as well as the pragmatic risk-avoiding strategyforeground the impact of the mediator (the translator) on the mediated discourse.However, in Halverson's cognitive explanation, the translator might not even be aware of these tendencies, whereas the pragmatic explanation offered by Pym has to do with conscious decision-making. 4An additional difference between them is that Pym discusses the question in exclusively translation-specific terms, whereas Halverson endeavours to bridge between translation and other metalinguistic activities, such as second language acquisition (Kuusi 2011: 170).This connection with other linguistic activities is central to drawing parallels between translation universals and the nature of discourse presentation.Chesterman (2004) proposes a consideration of translation universals as part of, and in the context of, other constrained communication, such as reporting discourse, or communicating in a non-native language (Chesterman 2004: 45).According to this view, universals of translation might not be restricted to translation alone, but characterize language mediation in general, and translation as one of its forms.This idea was first formulated by Blum-Kulka (1986), who, putting forward what she termed the explicitation hypothesis, suggested that translationboth professional and non-professionalas well as language produced by language learners display an increase in explicitness that cannot be traced back to differences in the languages in question (Blum-Kulka 1986: 19, 21).
Chesterman's suggestion has been taken up by Ulrych (2009), whose findings on mediated discourse provide evidence for all the aforementioned translation universals.Ulrych's data comprises different modes of mediation, such as translation and editing, produced predominantly in the context of the EU (Ulrych 2009: 223, 227-228).Her analysis provides evidence for the tendencies analyzed under the heading of translation universals in all these forms of language mediation (ibid.223, 227).Consequently, the observed regularities appear not as something translation-specific, but rather fall under the umbrella concept of mediation universals (ibid.223).Similarly to Ulrych's approach, Halverson (2003) verifies the results of her analysis concerning translation universals against research on second language acquisition.Halverson's study suggests that simplification and normalization (standardization) result from asymmetries in the cognitive organization of semantic networks (Halverson 2003: 221).Moreover, Halverson finds evidence for the influence of the same cognitive factors both in translations and in learner language alike (ibid.225).In her view, the cognitive constraints that result in simplification and normalization in translation seem to produce similar effects in learner language, too (Halverson 2003: 227).
Rather than characterizing translation alone, translation universals may, perhaps, be regarded as the translation-specific manifestations of the larger phenomenon of mediation universals.In the next section, I follow this line of enquiry by undertaking an analysis of narratological and literary-theoretical accounts of discourse presentation, and drawing parallels between them and the posited universals of translation.

Reportive interference as typification and schematization
As early as the 1930s Vološinov posited the inevitable impact of the mediating agent on the mediated discourse, emphasizing that all speech reporting includes active reprocessing of the 'Other's speech' (čužaja reč') (Vološinov 1973: 116-117;in Russian 1972in Russian [1930]]: 114-115; see also Bakhtin 1987Bakhtin [1975]: 340, 357-367)]: 340, 357-367).This idea has later been developed by other researchers, including some theorists outside linguistics (often without reference to Vološinov or Bakhtin).In literary theory, Sternberg (1982) discussed it as reportive or interpretive interference and considered it an intrinsic component of discourse presentation (Sternberg 1982: 85, 89).According to Sternberg, in discourse presentation the mediating agent's point of view inevitably affects the mediated discourse, making a simple reproduction practically impossible and leading to various modifications of the original discourse.Even direct discourse, which is conventionally assumed to be a faithful reproduction of a prior discourse, is always coloured by interpretive interference (ibid.70-74).The impact of the reporter's viewpoint is further enhanced by the communicative pressure present in discourse mediation: the communicative need to be understood leads to modifications of the original discourse and therefore to reportive interference (ibid.89).The degree of this interference varies, but the reporter's presence subjects all forms of discourse presentation to some degree of perspectival ambiguity between the voice or viewpoint of the reporter and that of the reportee (Sternberg 1982: 77, 92).Sternberg (1982) discusses discourse presentation as recontextualization, where the reporting frame, the new context of the displaced quote, "colors and comments" on the reported discourse (Sternberg 1982: 72).Recontextualization is also the term used by Ulrych (2009) in translation studies to describe different forms of discourse mediation.The two approaches, the one literary-theoretical and the other translatological, are strikingly similar: Sternberg sees recontextualization as interference caused by the reporter's point of view, while Ulrych (2009: 224) describes recontextualization as refraction, involving a change in perception.
The nature of this reportive interference is elaborated in post-classical narratology by Fludernik (1993).In her theory of schematic language representation, Fludernik (1993: 398) claims that all represented discourse is typical and schematic in nature.Rather than a mimetic reproduction of the original, represented discourse constitutes a typified representation of it (ibid.399).This is true even for direct discourse, usually associated with a verbatim reproduction (ibid.408).According to Fludernik, both direct and free indirect discourse can be used to typify and even misrepresent the original discourse (ibid.423).This is the case when the discourse represented is, for example, hypothetical or speculative, overtly acknowledged as fictional, or a condensed version of the original discourse (Fludernik 1993: 398-418).In oral narration, typified responses are often reported in the form of direct discourse, with the schematized nature marked by like as in "He's like 'You know, I really don't want to do this'" (ibid.416).The original discourse is not quoted word for word, but modified to fit the new context, to produce a discourse that meets the expectations of the new target audience and suits the speaker's aims.Instead of faithfully reflecting the original discourse, discourse presentation evokes some impression of the prior discourse, an impression that serves the reporter's communicative purposes (ibid.399).The purpose of discourse presentation is to create an impression rather than report faithfully, and in most situations, it is more important to present the gist of the prior discourse rather than its exact wording (ibid.426).Therefore, rather than a quotation or a copy, represented discourse presents a typical and conventional version of the original discourse.As Fludernik concludes, typification is a normal practice in discourse presentation, not an exception (ibid.418).Fludernik (1993) also points out that even the expressive and subjective features in mediated discourse (interjections, exclamations, intensifiers, evaluative adjectives etc.) need not be truly mimetic.Rather, they function as linguistic markers of subjectivity, a trigger mechanism that makes the reader attribute the discourse to a subjective consciousness, separate from that of the narrating agent (Fludernik 1993: 425, 428-429, 437).In a sense, subjective expressions are often used as 'enquotation devices' that, in absence of quotation marks or a reporting clause, indicate the presence of a subject of consciousness other than the narrator, but with the advantage of not implying a verbatim quote (Jahn 1994).
Both Sternberg and Fludernik present overwhelming evidence for the reportive interference present in discourse presentation, focusing especially on direct discourse and the 'reproductive fallacy' connected with it (Sternberg 1982: 89-92, 109-111;Fludernik 1993: 409-414; see also Jahn 1994).However, even if their arguments seriously undermine the idea of direct discourse as a faithful and mimetic quotation, this should not make us abandon the notion of faithfulness altogether.As Short et al. (2002) argue, the notion of faithfulness associated with direct discourse is a context-sensitive one, being conventionally more strongly associated with some contexts of discourse presentation than with others (Short et al. 2002: 327).For example, in newspapers, law courts or academic citation, the citations presented as direct discourse are expected to be faithful and accurate, whereas in contexts such as spoken discourse and fictional dialogue the same degree of faithfulness is not assumed (Short et al. 2002: 327-331).Although Short et al. do not discuss translation, they do nevertheless mention that one of the factors increasing the expectation of faithfulness is the reader's ability to verify whether the quotation is accurate or not (ibid.327).Translation, defined as free direct quoting in Mossop (see section 3), is conventionally expected to be accurate and faithful, and the pressure of quoting accurately is made that much more intense by the verifiability of said quotation.This puts translation in rather an awkward position.On the one hand, translation as a form of discourse mediation is never a verbatim reproduction of the original, and the mediating agentthe translatorhas an inevitable impact on the discourse mediated.On the other hand, by convention, readers expect translation to be a faithful presentation of the original discourse.
Fludernik's account underscores the nature of reportive interference as typification and schematization, based on the prototypical, the conventional and the recurrent.This typifying tendency in discourse mediation, suggested by Fludernik, seems curiously similar to normalization or standardization, proposed in translation studies as potential universals of translation.Similarly, the tendency to normalize or standardize brings to mind another concept from literary theory, namely that of naturalization.This term was coined by Culler (1997Culler ( [1975]]) to refer to an interpretive strategy on the part of the reader, exercised when encountering anything strange or seemingly illogical in the text.In such cases, the reader endeavours to interpret inconsistencies in line with her/his previous knowledge, that is, s/he attempts to make familiar that which seems strange, and render conventional that which seems odd in the text (Culler 1997(Culler [1975]]: 137-138).The concept of naturalization is thus a heuristic one, accounting for those strategies utilised by the reader to make sense of anything the reader experiences as inconsistent or incomprehensible (Culler 1997(Culler [1975]]: 137-138, passim).
The translator is also a reader, naturalizing the text in Culler's sense.When interpreting the text in order to solve and explain any inconsistencies encountered in it, the translator is applying a naturalizing strategy, be it on the level of minute linguistic detail or broad thematic idea.The crucial difference between a translator and any other reader lies in the fact that whatever interpretation the translator arrives at, it is likely to influence his/her translation.The outcome of the translator's naturalizing strategy is thus inscribed in the translation, and is likely to influence the way the translation will be interpreted by its readers.When re-framed in terms of translation theory, the naturalizing strategy, seeking to make the text familiar, normal and conventional, is likely to lead to standardization.Standardization is the consequence of the translator's attempt to make the text more natural.In Venuti's (1998) terms, we might call this attempt to naturalize a tendency towards fluency and domestication, a tendency that Venuti strongly disapproves of (ibid.6-7).Due to the contentious connotations prevalent in Venuti's view on domestication, I prefer the term standardization, used in translation studies in a more neutral sense.
Whether the strategies of naturalization and typification are conscious and deliberate, or whether they are unconscious and unreflected, is difficult to tell.Be that as it may, these strategies are likely to make a difference.If naturalization/typification is a conscious strategy, its influence on translation can be monitored and regulated by the translator.Unconscious tendencies, on the contrary, may have uncontrolled influence on the translator's output.

Mediation of discourse presentation: the case of free indirect discourse
In this section, I explore the workings of discourse mediation in the translation of free indirect discourse.Drawing on the results of my earlier research (Kuusi 2011), I analyze passages containing FID selected from Fedor Dostoevsky's Crime and Punishment and its Finnish translations.The translations are compared with their originals and with each other, and the changes made in them are approached as instances of reportive interference, typification, naturalization, explicitation and normalization/standardization.Free indirect discourse is a mode of discourse presentation recognized for its ability to interweave the point of view of the reporter (the narrator) with that of the reportee (the fictional character) in the same utterance (see Pascal 1977).Hence, FID highlights the simultaneous presence and the blending of two voices, the mediating and the mediated, in a single utterance.Even though the two viewpoints are simultaneously present in all forms of discourse presentation, ID and DD are normally not considered multi-voiced.In FID, in contrast with ID and DD, the duality of voices and perspectives is linguistically more salient as well as generally acknowledged (Sternberg 1982: 70).
The (constructed) examples 1-6 below illustrate how the forms of speech representation either separate or merge the mediating and the mediated discourse.
(1) "Am I doing the right thing?" she thought.
(2) She wondered whether she was doing the right thing.
(3) Was she doing the right thing?(4) Lord, was she really doing the right thing?
(5) "Lord, am I really doing the right thing?" she thought.
(6) Lord, am I really doing the right thing?
In direct discourse (1), the two voices are clearly separated.The reported discourse retains its original pronouns, tense and other deictic expressions, and is clearly separated from the reporter's discourse by quotation marks.In indirect discourse (2), the reported discourse is completely controlled by the reporter's discourse: the back-shift of tenses and the substitution of the first-person pronoun with the third clearly indicate the reporter's deictic orientation.In addition, the inverted word order is substituted by an indirect question without inversion, indicating the syntactic subordination of the reported discourse to the reporting one.In free indirect discourse (3), the voices of the reporter (the narrator) and the reportee (the fictional character) merge.The third-person reference to the subject of consciousness and the use of the past tense indicate the narrator's presence in and control of the discourse, but the voice and point of view of the character surface through direct question.This impression is intensified by the semantic content of the discourse that is likely to be attributed to the character rather than the narrator.However, the omission of the reporting clause renders implicit the attribution of the discourse to a character, leaving it up to the reader to draw such a conclusion.Therefore, as the FID in (3) does not contain clear expressive features signalling the subjectivity of the character, it can also be read as pure narrative report, in cases where the context allows such interpretation.The experience of the character can also be more clearly indicated in FID by expressive and subjective features as demonstrated in (4).These features (Lord, really), characteristic of direct (5) and free direct discourse (6), are normally omitted in indirect discourse (2) but not in its free counterpart.As predicted by Fludernik's schematic language representation (see section 5), the expressive features in (4) clearly indicate the presence of a subjective consciousness different from the reporting one, serving as a 'schematic indication of alterity' in Fludernik's sense (1993: 437).(For a general overview of direct, indirect and free indirect discourse, see, for example, Lehtimäki & Tammi 2010; for a more detailed treatment of FID, see McHale 1978;Leech & Short 1981.)On the linguistic level, then, FID accommodates features of both narratorial and character discourse in one sentence, thus inviting the reader to construct both an impression of the narrator's controlling presence and the character's surfacing subjectivity.The inclusion of features from both direct and indirect discourse makes the linguistic form of FID incoherent on the linguistic surface.The implicitness of the reported nature of the discourse (no reporting clause) makes FID an elusive mode, an interpretative rather than a strictly linguistic category: consider, for instance, example (3), permitting a reading both as FID and as narration.As frequently noted in narratological accounts on FID, the mode cannot be defined exhaustively in strictly linguistic terms, and therefore functions as a fuzzy concept (see, for example, Fludernik 1993: 437).However, notwithstanding the implicit quality in FID, in most cases the reader has no difficulty in identifying the character's voice in cases such as (3) and (4).
There are several studies on the translation of FID, reporting on the substitution of the FID of the original with DD or ID in translations.These studies report on literary translations from the 17 th to the 20 th century between different language pairs: from English into Finnish, French, German, Hebrew and Russian; from Russian into English, German, French and Finnish; from French into German; and from Finnish into Russian (see, for example, Levenston & Sonnenschein 1986;Kittel 1990;Jekutsch 1995;Rouhiainen 2000;Bosseaux 2004; for a more detailed account, see Kuusi 2011: 113-129).
In Kuusi (2011), I compared passages of FID from Dostoevsky's Crime and Punishment with their translations into Finnish (dating from 1888-1889, 1907, 1922, 1970, 1986and 2008) and English (1914, 1992)).In that analysis, FID and other modes of discourse presentation were employed as tools for identifying the narrative point of view in both the source and the target text and for locating and categorizing translation shifts in point of view.Linguistic changes accompanying these shifts were then analyzed and classified.The results of this analysis were in line with earlier research on the translation of FID: in all but one of the 24 passages analyzed, FID was lost or weakened in at least one of the eight translations, and in 17 of 24 passages FID was lost or weakened in at least four of the translations.Additionally, my analysis also revealed that all linguistic changes resulting in the weakening or loss of FID qualified as manifestations of explicitation and normalization -something that had gone unnoticed in earlier research on the translation of FID.The most typical amongst them were the addition of a reporting clause and the replacement of the third person pronoun referring to the subject of consciousness with either a first person or a noun reference.Examples ( 7) and ( 8) below demonstrate the loss of FID in translations, as well as the resulting standardization of discourse presentation.In (7a), narration is followed by FID, when description of external facts is succeeded by thoughts occurring to Luzhin (a rather unpleasant character in the novel, who is a petty and presumptuous clerk.Luzhin wants to marry a poor girl like Dunya Raskolnikova, so that she might be forever indebted to him).The narration of the first two sentences introduces the thoughts of the character to the reader, but in the third sentence the mode of presentation switches to FID, expressing a conclusion the character himself arrives at.Following the (irregular) 19 th century convention, FID is placed in quotation marksa convention that in Dostoevsky's use of FID indicates a move from semi-conscious thought to more conscious reflection (Pascal 1977: 127, 129;Kuusi 2011: 53-54).
(7a) С болезненным ощущением припоминался ему, тоже как-то невольно, Разумихин... но, впрочем, он скоро с этой стороны успокоился: "Еще бы и этого-то поставить с ним рядом!" Но кого он в самом деле серьезно боялся, -так это Свидригайлова... (Dostoevskij [1866(Dostoevskij [ ]/2006: 284) : 284) 'With a painful feeling he remembered, also somehow involuntarily, Razumikhin ... however, he soon set himself at ease in that regard: "As though that [person] too could be held up to him!"Indeed, who he was really afraid of, -it was Svidrigailov ...' 5 (7b) Kipeänä aistimuksena ja myös kuin vastentahtoisesti hänen mieleensä muistui myös Razumihin... mutta tämän suhteen hän rauhoittui pian: "Voisiko muka sellaisen asettaa minun rinnalleni!" Mutta se, jota hän aivan tosissaan pelkäsi, oli Svidrigailov... (Dostojevski 2008: 387) 'With a painful feeling and also somehow involuntarily he remembered Razumikhin ... however, he soon set himself at ease in that regard: "As though [a fellow like] that could be held up to me6 !"But the person that he was seriously afraid of was Svidrigailov ...' 7In example (7b) from the new Finnish translation by Kuukasjärvi ( 2008), the third-person pronoun referring to the subject of consciousness has been replaced with the first-person pronoun.In the original, the linguistic features of FID (personal pronoun) refer in part to the narrator, whereas other features (the exclamation and the quotation marks) indicate the discourse as belonging to the character.In (7b), the personal pronoun is in line with these other markers, and the uncommon combination of the original has been replaced with one that is (more) common.The change of personal pronoun affects the mode of presentation: FID turns into (free) direct discourse, the standard mode offering access to the protagonist's mind.Consequently, the dual-voice effect of the original FID is lost: in the translation of the sentence, only one voice remains, that of the character.As a result of this narrative change, when the narrator is not present in the quotation, the reader assumes the character takes full responsibility for verbalizing it.Thus the shift is likely to affect the reader's interpretation of the discourse, attaining a near-verbatim character.This is due to the fact that different modes of discourse presentation are accompanied by varying assumptions regarding faithfulness, and, despite the evidence provided in literary theory, the default interpretation in the case of direct discourse is a verbatim representation of the reported discourse (Short et al. 2002: 332;Fludernik 1993: 435;Pascal 1977: 131).A verbatim representation presupposes 5 The auxiliary translations in single quotation marks are based on the English translations of the novel by Richard Pevear and Larissa Volokhonsky (1992) and Constance Garnett (1914), in this case of Pevear and Volokhonsky: 'With a painful feeling he also somehow involuntarily remembered Razumikhin ... however, he soon set himself at ease in that regard: "This is the last person who could be held up to him!"Indeed, if there was anyone he was seriously afraid of, it was -Svidrigailov ...' (Dostoevsky 1993(Dostoevsky [1992]]: 308).a verbalized original, and verbalization, in turn, presupposes conscious reflection (see Cohn 1978: 139;Pascal 1977: 131).When read as a conscious thought -instead of some vague impression evoked in the character's mindthe discourse arguably makes Luzhin seem even more small-minded and presumptuous than in the original.
By shifting from the more unconventional FID to the more conventional direct discourse, the translation normalizes the original discourse, supporting the hypothesized translation universal of normalization (standardization).However, the shift, illustrating the interpretive interference present in all discourse presentation, can just as easily be described as typification in Fludernik's sense.It is likely to result from the naturalizing strategy of the translator, making sense of the unconventional combination of the third-person pronoun combined with the quotation marks, the exclamation and the content of the discourse.In fact, the replacement of the third-person pronoun with the first-person pronoun actually demonstrates that the translator has attributed the discourse to Luzhin, and not to the narrator.Judging by the change of pronoun, one can conclude that in spite of the third-person pronoun in the original FID, the impression of the character's voice has been strong enough to make the translator read the discourse as belonging to the reported consciousness.At the same time, the shift could also be motivated by the recipient-oriented nature of mediated discourse: if the translator understood the original form of presentation to be unconventional, s/he might have wanted to increase its readability by shifting to the direct mode.In summary then, in (7b), the notions of naturalization, typification and standardization describe the translation more accurately than the notion of free direct quoting.The translation is not direct in the sense Mossop (1998) intended: direct forms necessarily preserve the deictic orientation of the original discourse, but this has not been preserved here.In this sense, the faithfulness assumption has not been met, and in Mossop's (1998) terms the translation would be only marginally translational.As Taivalkoski-Shilov (see section 3) has noted, actual translations are not always direct in the sense suggested by Mossop.
The same change has been made in four of the six Finnish translations of the novel: the translations dating from 1922, 1970, 1986 and 2008 all shift to direct discourse.Only the first two Finnish translations, dating from 1888-1889 and 1907, retain the third-person pronoun and the use of FID.Interestingly, this runs against the retranslation hypothesis (see, for example, Paloposki & Koskinen 2004).This question, however, does not fall within the scope of this paper, and is dealt with elsewhere (see Kuusi 2014).The fact that the shift to direct discouse is not evinced in the first two Finnish translations indicates that the unusual combination of quotation marks coupled with the third person referring to the subject of consciousness may have brought about the change of the personal pronoun.The combination, however unusual it may be from a contemporary perspective, was not quite so unusual in the 19 th century.In the first two translations dating from the 19 th and the early 20 th century, the combination did not confuse the Finnish translators; later in the 20 th century, however, the translators may have found it unconventional and thought it detracted from the readability of the text.
Example (8a) represents in FID the thoughts of Raskolnikov, the main character of the novel, whilst on his way to meet Svidrigailov.Wondering how Svidrigailov could possibly help him in his situation, Raskolnikov suddenly thinks of Sonya.In this context, the mention of her name with a question mark is enough to convey the impression that this is a spontaneous thought on the part of Raskolnikov.Even without attributing the thought explicitly to the character, the expressive featuresthe questions, the modal expression может быть 'perhaps', the incompleteness of the first sentence, the intensifier да и 'and besides', and the temporal expression теперь 'now' all function as "indications of alterity" in the Fludernik sense, thus evoking the impression of a subjective consciousness.
(8a) Соня?Да и зачем бы он пошел теперь к Соне?Опять просить у ней ее слез?(Dostoevskij 2006(Dostoevskij [1866]]: 429) 'Sonya?But why would he go to Sonya now?To ask her for her tears again?'8 In two of the six Finnish translations, those dating from 1922 and 1986, the translator has added a reporting clause hän ajatteli 'he thought of' and omitted the question mark.On the one hand, the reporting clause confirms that the thought belongs to the character, and not the narrator.On the other hand, it transforms the thought occurring to Raskolnikov into a narrator's report of the thought in the form of ordinary indirect discourse.For a moment, the impression the reader has of direct access to the mind of the character is lost, and the dualvoice quality of FIDthe merging of the voice of the reporter with that of the reported speakeris lost (Kuusi 2006: 97-98).The reader is momentarily distanced from the fictional character.
(8b) Hän ajatteli Sonjaa.Mutta miksi hän nyt menisi Sonjan luo? Jälleenkö anelemaan häneltä säälin kyyneliä?(Dostojevski 1964(Dostojevski [1922]]: 447-448) 'He thought of Sonya.But why would he go to Sonya now?To beg her for mercy tears yet once more?' The addition of a reporting clause is an instance of explicitation, i.e. of clearly expounding something that was only implicit in the source text.However, this could also be construed as an instance of normalization, since an incomplete sentence has been replaced with a complete one (Kuusi 2011: 156-157).As Pym (2008) observes, "if one reduces the reader's options by taming ambiguity, the effect must be greater standardization" (Pym 2008: 314-315).In this case, explicitation is normalization.Here it is helpful to remember that in Pym's treatment of translation universals (see section 4), both explicitation and normalization fall under the general tendency to standardize.
In three of the six Finnish translations of this passage, translators have supplemented their translations with additional information.In (8c) from the 2008 translation, the translator has expanded the question by adding Entä 'And what about' in the beginning of the sentence, but preserved the free form (no quotational phrase) and the question arising in the character's mind.The same change has been made in the 1970 translation.In (8d) from 1907, the question has been expanded by adding a modal expression Ehkä 'Perhaps'.The additions in (8c) and (8d) do not affect the mode of discourse presentation or alter the impression of hearing the character's own voice.They can hardly be thought of as rendering the text more explicit, but rather expand upon the text somewhat.The changes made in (8c) and (8d) do not distance the reader from Raskolnikov like those in (8b).Rather, these changes enhance the impression of the character's subjectivity by adding a modal expression of speaker uncertainty (8d) and prefacing the question in a way that emphasizes the occurrence of a new thought (8c).Examples (8c) and (8d) demonstrate that not all changes that expand the discourse necessarily lead to changes in mode of discourse presentation.One need not preserve the exact linguistic features in order to preserve FID.Changes do not affect the mode of discourse presentation, if they do not shake the balance between the reporting and the reported discourses in the original.
(8c) Entä Sonja?Minkä tähden hän nyt menisi Sonjan luo? Taasko anelemaan tältä säälin kyyneleitä?(Dostojevski 2008: 580) 'And what about Sonya?But why would he go to Sonya now?To beg her for her mercy tears yet again?' (8d) Ehkä Sonja?Mutta miksi menisi hän nyt Sonjan luo? Ehkä taas kerjätäkseen tämän kyyneleitä?(Dostojevski 1907: 533-534) 'Perhaps Sonya?But why would he go to Sonya now?Perhaps to beg her for her tears again?' In (8b) we can observe tendencies similar to those in (7b).In (8b), the more unconventional mode (FID) is substituted with the more conventional one (indirect discourse).The translation renders explicit and normalizes the discourse, providing evidence for standardization or, in Fludernik's terms, typification.The reportive interference results from the translator's naturalization of the text: the addition of a quotational phrase confirms that the translator has read the incomplete sentence as a thought appearing in the mind of the character.However, if the decision to add a reporting clause has been consciously taken, it is likely to have been motivated by the translator's pragmatic considerations concerning the reader and the readability of the text.The translation is not free in the Mossop sense, because a quotational phrase has been added to describe the speech act.Thus, in both (7b) and (8b), we have evidence of standardization, a possible universal of translation; of typification of discourse presentation; and of the naturalizing interpretive strategy of the translator.
Even though examples (7b) and (8b) provide evidence for standardization, the tendency to standardize might not affect all translations in exactly the same manner, nor are all forms of discourse presentation necessarily treated in the same way in translation.As noted above, in FID the voices and points of view of the reporter and the reported merge, and this qualitythe duality of voicemanifests itself in the linguistic form of FID.In FID, there is a seemingly incoherent mixture of features traditionally associated with direct discourse and those associated with indirect discourse.As I have suggested elsewhere (Kuusi 2011: 77-79, 171-172), this incoherent linguistic surface of FID, together with its implicitness in discourse attribution, seems to call for a normalizing or typifying intervention of the mediating agent.It is the nature of FID, then, that might be partly responsible for the standardization observed in the different translation versions we have just examined.Therefore, we may expect the translation of other modes of discourse presentation to exhibit the same tendencies to a lesser extent.
Standardization, resulting from the naturalizing strategy of the translator, may be conscious and recipient-oriented in nature, but it may also be unconscious.Naturalization as an interpretative strategy must necessarily arise from the reader's mind, from his or her cognition, and therefore constitute a cognitive explanation.Regularities of mediated discourse, on the other hand, are thought to arise from the need to meet the target reader's expectations (Ulrych 2009).Whereas reader-oriented reasons are necessarily conscious, the cognitive ones do not have to be.In translating FID, both may have their own parts to play.The translator's cognitive processingwhether conscious or notnaturalizes the textual incoherencies so as to arrive at a coherent interpretation.After such an interpretation has been reached, the translator may anticipate the difficulties encountered by the reader and try to unburden the reader of the discourse sense-making task.
Both explicitation and normalization have been described in translation studies as tendencies that aim at increasing the readability of a text.Readabilitythe quality of a text that makes it easier for a reader to processconsists of conventionality (answering the reader's expectations) and sufficient cohesion, in part realized by an appropriate level of explicitness in the text.These are the goals that the translators of (7b) and (8b) might have pursued, without noticing the resultant simplification of narrative voices and perspectives.The change of the pronoun in (7b) may have been conscious in and of itself, but the subsequent change in the mode of discourse presentation might still have been unintended.Therefore standardization, even when meant to increase the readability of the text, may have other, unintended effects on the reading experience.
As Charlotte Bosseaux (2004: 110, 121) notes, regular shifts in the narrative point of view of the translation may alter the overall impression or "feel" of the text.To understand how the loss of FID affects the reading experience, one needs to understand why FID was employed in the narrative in the first place.The general functions of FID are often best described in comparison with those of ID and DD.Unlike ID, FID retains the immediacy and vivacity of DD, but unlike DD, in FID the thoughts and experiences represented need not be fully conscious and verbalized (Pascal 1977: 131-132, 137;Cohn 1978: 103).Like ID, FID conveys the narrator's attitude towards the discourse represented, albeit in an implicit way.This attitude is sometimes ironic, distancing the reader from the character, and at other times it is empathetic, creating a feeling of closeness with the character (McHale 1978: 275).Leech and Short (1981: 346-347) observe that modes of discourse presentation can be employed by novelists to control the readers' sympathies towards the characters: the readers are likely to feel more for those characters whose discourse is presented in FID.The results of the psycho-narratological experiment conducted by Bortolussi and Dixon (2003: 230-236) provide strong evidence for the influence of the mode of presentation on the reception of a text.In an experimental study, Bortolussi and Dixon manipulated the way in which the characters' discourse was presented in the story, and analyzed the attitudes of their test readers towards fictional characters whose speech was presented either in FID or DD.The results showed that the mode of presentation had an effect on the readers: the readers regarded the character whose speech was presented in FID as more rational.
In addition to its general functions, FID may also have functions specific to a certain narrative.As I have argued elsewhere (Kuusi 2011: 109-110;2014: 139-140), the use of FID is an important feature in Dostoevsky's novels, foregrounding their polyphonic and dialogic nature.However, it is not within the scope of this paper to fully explore the question of how changes such as those in (7b) or (8b) affect the reading experience.Suffice it to say that whenever FID is weakened or lost in translation, the functions assigned to it in the original are also likely to be weakened or lost.In both cases, the narrative point of view is altered, and the merging (or clash) of voices present in FID is subdued or replaced by the unity of voice.

Conclusions
In translation, FID is often replaced with more conventional modes of discourse presentation, such as direct or indirect discourse.This can be regarded as evidence for standardization, a potential universal of translation.However, it has been reported that mediated discourse, such as editing or non-native text production, may display the same standardizing tendencies (Ulrych 2009: 229).Consequently, translation universals no longer appear as translation-specific, but rather as typical of all discourse mediation in general.
The connection between translation universals and the nature of discourse presentation (reported speech) has received less attention so far.The translation universal of standardization, alternatively referred to as normalization or conventionalization, bears a clear resemblance to the mechanisms of typification and schematization that, according to Fludernik (1993), are present in all speech and thought representation.At the same time, the reasons behind standardization, residing in human cognition, may be linked to the concept of naturalization (Culler 1997(Culler [1975]]), referring to the reader's strategy of reconciling anything that seems strange or inconsistent in the text.As a heuristic concept, the notion of naturalization operates on a different level with translatological research on cognitive factors underlying translation universals (see Halverson 2003).Still, the notion may be useful in explaining the translator's motivation behind standardization.
As discussed above in section 3, Mossop (1998) finds directness an important defining feature of translation: of the different potential modes of discourse presentation, he draws a parallel between translation and direct discourse, described by Sternberg (1982: 69) as "the product of a tug-of-war between representational fidelity and communicative demand".Even if Mossop seems to relegate directness to preserving the original first-person forms (1998: 251), none the less, an idea very similar to that of Sternberg is expressed in Mossop as the "duality of translational quoting" (1998: 253); in Mossop's view, translation is both source-and target-oriented in its reporting of something and to someone, and therefore "an inherently Janus-like concept" (ibid.).Here, the descriptions of translation and direct discourse presentation definitely align.
Despite the parallels between translation and discourse presentation, there is a fundamental difference between the two.Unlike discourse presentation in fiction, translation is a form of discourse mediation with a real and concrete source text.In fictional discourse presentation, the prior discourse is normally non-existent; it is something that the reader constructs on the basis of the represented one (see Sternberg 1982: 112).In translation, the source text is there for the translation to be checked against or compared with.Consequently, observing the regularities of translational behaviour can help us to concretize the impact of an act of representation on represented discourse, as discussed in literary theory.At the moment, the observations made in research on translation universals seem to support the narratological and literary-theoretical notions concerning the nature of discourse presentation, namely reportive interference, naturalization and typification/schematization of the reported discourse.Thus, the interdisciplinary approach applied in the present article can be mutually beneficial: just as narratological and literary-theoretical accounts of discourse presentation offer new insights for the study of literary translations, studying translations may deepen our understanding of the mechanisms of discourse presentation in literary narratives.