Inglês (pdf)
Artigo em XML
Referências do artigo
Tradução automática
Enviar este artigo por email
Citado por SciELO 
Similares em
SciELO 
Permalink
1. Introduction
One of the greatest sources of enjoyment in reading Jane Austen’s Mansfield Park is the juxtaposition of discourses contributing to characterization, from its heterodiegetic narrator to the characters. The attribution of a source can often be ambivalent, in part due to Austen’s “free mixture of different modes of speech presentation” (Page, 1972/2011, p. 123), a refinement of resources already present in the writings of other 18th century authors (Spencer, 2009). Bray acknowledges that “Throughout Austen’s fiction the subtle interplay between these categories of speech representation is a vital means of characterization” (Bray, 2018, p. 34). As the plot advances and characters’ principles are put to the test in every decision they take, dialogue scenes are presented to the reader as a source of “internal evidence” for the reader to judge the “moral and social nature of characters” (Morini, 2009, p. 9).
Austen recognized that dialogue-writing posed a technical problem, namely “how to give the impression of a sequence of speeches without such loss of narrative pace as might be damaging to the rhythm of an episode” (Page, 1972/2011, p. 121). Of all speech presentation forms, economy of expression can be achieved particularly through Narrative Report of Speech Acts (Leech & Short, 2007). In (Semino and Short, 2004, p. 12), NRSA is defined as the mode through which “the speech act value of the utterance presented is indicated, often with a specification of the topic of the speech act, but no more elaboration of what was said in the anterior discourse is made”. 1 NRSA may lack a topic, in which case it has the function of “announcing and introducing utterances which are then reported in more detail in the following text” (Semino & Short, 2004, p. 75).
In the discourse structure of fiction as conceived in Leech & Short (2007, pp. 210-218), with different embeddings of communicative acts, the SA (Speech Act) in NRSA concerns the communicative situation at the character level imagined by the narrator, while the R (Report) concerns the narrator reporting it in communicative situation shared with the reader. NRSA thus contains information about two addressers, the character who speaks and the narrator who reports, while granting the latter almost full control, because the claim to faithfulness to the ‘original’ utterance (the character’s) is lower than in any other speech reporting forms: “the narrator does not have to commit himself entirely to giving the sense of what was said, let alone the form of words in which they were uttered” (Leech & Short, 2007, p. 323). In this model of embedded discourse levels, each of the addressers is a ‘voice’ allowed to express his or her attitude and stance towards the contents being uttered, in particular by reference to the values shared with the addressees. On the character level, NRSA condenses information about the characters’ intentions and about their stance, offering cues for the reader to infer personality traits and to construe a “value picture” of each character (Leech & Short, 2007, p. 220). On the narrator-reader level, NRSA offers information on the narrator’s stance, which needs not coincide with the character’s, giving rise to a complex value pictures and ironical readings. In Mansfield Park it is notoriously difficult to assess the narrator’s stance because Austen weaves a “web of ‘evaluative opacity’” (Morini, 2009, p. 65) by means of unexpected lexical choices breaking an apparently predictable flow of discourse, i.e., by contrasting character stance with narrator stance.
The question addressed in this paper is how the lexical choices made by Austen in her instances of NRSA contribute to the reader’s potential interpretation of narrator and character intentions and stance. The proposal put forward is to analyse each of the levels in the discourse structure of fiction combining a speech-act theory analysis of the character’s speech with an analysis of both the character’s and the narrator’s dialogic stance.
Austen’s reception in Europe has historically been mediated by translation (for an overview, see e.g., Mandal, 2012 and Owen, 2018). Translation adds a new layer to the literary discourse situation, or a new link in the communicative chain, so that “the [implied] translator shapes the narrator” (Winters, 2010, p. 165, based on Schiavi, 1996) and has the potential to alter the signature resources of an author’s style. 2 Literary translators face two challenges: as readers of the original, they must identify all the concurrent interpretation possibilities that lead to the impression of opacity. As translators, they must attempt to reproduce them in the target language, i.e., re-instantiating the original text in its whole potential (Farias de Souza, 2013, p. 580).
Research on translations of Austen’s work into European languages focusing on speech and thought reporting forms has provided evidence of a tendency to disambiguate the narrator’s vs. characters’ voices, with effects on the way the readers access both. For example, several studies have found a tendency to translate Free Indirect Speech (FIS) passages as either Direct (DS) or Indirect Speech (IS) in Catalan (Alsina, 2008), Portuguese (Rosa, 2009) and Spanish (Alsina, 2011; Zaro, 2006, Green, 2017, Kiezelbach, 2017. Translation of FIS into DS enhances the character’s control of the speech being reported, while translation of FIS into IS enhances the narrator’s control (for the notion of control, see Leech & Short, 2007, p. 260). Such observations raise the question whether Austen’s NRSA is subjected to similar shifts in narrative form, particularly when the reporting verbs are used for evaluation purposes.
The present study aims to formulate an integrative analytical approach to NRSA and to apply it to a sample comprising the original and two sets of translations. On the one hand, the study hopes to contribute to the study of NRSA; on the other, the study aims to observe the extent to which the semantic density of NRSA has been conveyed. The sample consists of 9,424 words from chapters 1, 4 and 5 of Mansfield Park and its translations into Spanish and German published in the last two decades of the 20th century. Chapters were chosen because of the role of specific conversational interactions that are presented as NRSA in the development of events. Chapter 1, amply commented in the literature (see, e.g., Doody, 2012, pp. 175-176; Morini, 2009, pp. 65-74), provides background information on the Price and Bertram families and reports the decision process to bring Fanny Price to Mansfield Park (including discussions involving Sir Thomas, his wife Lady Bertram, and her sisters Mrs Price and Mrs Norris). Chapters 4 and 5 report the introduction of Mary and Henry Crawford, with “gay and brilliant introductory scenes” (Wiltshire, 2006, p. 61), and new interactions between them and the Park’s inhabitants and acquaintances. The protagonist character Fanny Price is attributed very few speech acts and only one, barely audible, with any illocutionary point in Chapter 5. 3
The chronological choice of translations is grounded on the comparability of the sociohistorical contexts. The Spanish translations were published in Spain in 1995 and 1996, respectively. As recorded by (García Soria, 2015), the translator of SP1 (Spanish Translation 1) is Miguel Martín Martín; the work was first published by Rialp in 1995. It is the translation with the widest circulation as it was reprinted by Plaza y Janés, later in their DeBolsillo (literally, ‘pocket’ book) imprint, and since 2004 Planeta has been publishing it in its series “Colección de Novela Romántica” (Romantic novel series). In 2015 Penguin used it to publish a Commemorative edition in its “Penguin Clásicos” (Penguin Classics) series, with an introduction by the scholar Tony Tanner. The translator of SP2 (Spanish Translation 2) is Francisco Torres Oliver, Spanish Translation Prize laureate in 2001. It was first published by Alba Editorial in 1997, as the opening book for the “Alba Clásica” (Alba Classic) series.
The German translations are a decade older. GT1 (German Translation 1) is by Christian and Ursula Grawe ([1984/2010) and was published by Philipp Reclam. According to (Bautz, 2007), “Reviewers describe the [Grawe] translations as ‘exceedingly reliable and stylistically adequate’”. GT2 (German Translation 2) is by East German translator Margit Meyer (1989/2010); in the late eighties, her translation was published by Aufbau-Verlag after revision by editor Klaus Udo Szudra. In recent years new translations into both languages have been published: in German by Manfred Allié and Gabriele Kempf-Allié (Austen, 2017), and in Spanish by Miguel Ángel Pérez Pérez (Austen, 2013). Differences between these and the late 20th century versions are left out of the scope of the present study.
The structure of the paper is as follows. Section 2 presents the frameworks used in the analysis of speech reporting expressions as well as the basic theoretical assumptions adopted for the analysis of translations. Section 3 presents the methodology of the study based on the identification and classification of Source text-Target text segments. Section 4 presents a general overview of NRSA in the sampled chapters and of the translation solutions identified in the four translations. Section 5 focuses on the types of shifts in illocutionary features and engagement value. Finally, section 6 is devoted to a final assessment of the model.
2. Preliminary definitions
2.1. The Speaker perspective: Speech Acts and illocutionary features
The study of fiction has benefitted from the concepts and methods of research on naturalistic speech (for a thorough overview see, e.g., Locher & Jucker, 2017). As Morini argues, “if a literary dialogue is to be effective, the reasons and drives motivating characters’ behaviour must be comprehensible to the reader; that is, they must be roughly the same as those motivating real people’s behaviour” (Morini, 2007, p. 6). In Pragmatics, Grice’s Cooperative Principle and Politeness Theory support insightful accounts of Austen’s technique (Davidson, 2004; Morini, 2009), while Speech Act Theory “can be enlisted to provide a framework for an analysis of the dynamic power of language to act, to mould contexts, and, ultimately, to create social constructs” (Nolan-Grant, 2009, p. 863).
The concept of Speech Act in (Austin, 1962) highlights the functional nature of language and the importance of speaker intentions in communication; through every utterance the speaker performs at least a locutionary act and an illocutionary act; the speaker’s intention determines its illocutionary force. (Searle and Vanderveken, 1985) further refined the definition of illocutionary force as a set of features, namely illocutionary point, degree of strength, mode of achievement, content and preparatory conditions, sincerity conditions and the strength of sincerity conditions. These features are conditions that have to be verified when interpreting an utterance as the performance of a speech act. Illocutionary force is a source of cues for implicit characterisation, i.e., based on inference (Culpeper, 2001).
Reporting verbs can convey the illocutionary force of the reported utterance, a notion that is made explicit in one of the most generally accepted functional classifications of speech reports, Caldas-Coulthard’s taxonomy of “verbs of saying”. She distinguishes speech-reporting verbs, descriptive verbs and transcript verbs (Caldas-Coulthard, 1994, p. 306). The class of speech-reporting verbs includes neutral structuring reports that introduce the locutionary act without evaluating (e.g., say and reply), metapropositional reports, which indicate the illocutionary point of the speech act (e.g., urge and explain), and metalinguistic (e.g., narrate). Descriptive verbs, by virtue of expressing prosodic and paralinguistic features, allow to infer manner and attitude of the speaker (e.g., cry and laugh). Finally, the class of transcript verbs is not considered speech-reporting but marks the relationship of the reported speech act to other parts of discourse (e.g., hesitate and repeat). Although this taxonomy was developed for the study of the discourse of the press, it has been applied to stylistic analysis of characterisation cues, for example in (Ruano San Segundo, 2017), an exploration of Dickens’s novel Nicholas Nickelby. The metapropositional illocutionary and descriptive categories were found to “contribute to the portrayal of fictional characters” (Ruano San Segundo, 2017, p. 107), through cumulative usage. These fine-grained distinctions are relevant for a comparative analysis of source texts and their translations, since the translator’s lexical choices may change the degree of explicitness of illocutionary force intended by the author for the reader.
This is illustrated by the results of a case study involving a translation from English into Spanish (Bourne, 2002), which analysed 56 occurrences of the reporting verb say introducing an impositive directive speech act in their inquit formulae (i.e., “he said” introducing an utterance that is clearly a command). In two-thirds of the instances, the translation introduced lexical variation specifying different illocutionary points (15 different verbs). The author concluded that the selection of verbs specifying illocutionary force can influence the reader’s construal of a character’s personality (Bourne, 2002, p. 251). These results can be explained by the influx of norms and style conventions prevalent in the Spanish writing tradition, where lexical repetition is considered inelegant and monotonous. Such expectations regarding lexical variety and richness might lead to overtranslating illocutionary features that will interfere with authorial intention. As regards German, (Winters, 2007) suggests a similar account for the differences in lexical variation patterns of speech-act report verbs between two translations of Hemingway’s The Beautiful and Damned.
(Busse, 2020) applies Caldas-Coulthard’s typology in her classification of reporting verbs in her 19th-century narrative fiction corpus. There is, however, no study using it to describe NRSA, which is hoped to be one of the contributions of the present study.
2.2. The Addressee perspective: dialogic stance
In addition to speech acts and their conditions, utterances encode the speaker’s dialogic stance, “the speaker/writer’s interpersonal style and their rhetorical strategies according to what sort of heteroglossic backdrop of other voices and alternative viewpoints they construct for their text and according to the way in which they engage with that backdrop” (Martin & White, 2005, p. 93). This dimension of communication is referred to as Engagement in Appraisal Theory (Martin & White, 2005, pp. 99-100). A first-level distinction can be made between monoglossic and heteroglossic statements. Monoglossic statements present propositions as taken for granted, in different semantic configurations such as presuppositions and categorical bare assertions (i.e., non-modalised). In contrast, heteroglossic statements present propositions as not taken for granted; typical linguistic realisations are related to the systems of modality and evidentiality, and include modal verbs and adjuncts (mainly with epistemic meaning). Verbs of reporting play an important role in heteroglossic stance as well.
Within heteroglossic positioning, textual formulations can either close or open the space for alternative positions: the former are instances of dialogic contraction (Martin & White, 2005, p. 103); the latter are instances of dialogic expansion. Each type varies according to (a) the subjectivity that is presented as the source of the proposition (the addresser or a third person) and (b) the authorial alignment - or lack thereof (disalignment) - with the source. The grammar of reported speech and thought typically expresses different degrees of association between the text’s internal authorial voice and the external source. In dialogic expansion disassociation between them is expressed as attribution (Martin & White, 2005, pp. 111, 134). Depending on whether the authorial voice takes responsibility for the contents of the proposition, attributions may acknowledge (with verbs such as say or state) or distance (with verbs like claim). As far as alignment with the reported proposition is concerned, the authorial voice can endorse it (e.g., with prove) or distance itself from it (claim). The co-occurrence of endorsing and distancing formulations can be used to signal a clash in values, leading to irony (see Leech & Short, 2007, p. 223).
Features of illocutionary acts can be interpreted as evidence of the dialogic stance adopted by the speaker, particularly when he or she is the source of the proposition. Assertive acts with declare, assert and proclaim convey the speaker’s firm belief that something is the case, a higher degree of illocutionary strength than acts with say. This firmness of belief can be associated with a contractive dialogic stance towards the addressee, resulting in pronouncement, when the source is the speaker, or in endorsement, when it is a third person. Variations in strength can signal variations in stance; for example, the assertive speech act verbs suggest vs prove differ in illocutionary strength, and they also differ in stance, from acknowledgment (expansive) to endorsement (contractive). Stance is also a source of information about the power status and the solidarity of interlocutors; for example, since a contractive stance challenges, confronts or resists particular dialogic alternatives, the speaker must feel in a capacity to engage the addressee in such confrontational terms. A narrator reporting that a character is ‘urging’ a course of action is giving cues as to his or her social standing relative to the hearer’s.
As will be seen, translation shifts in the stance dimension can alter the original characterisation cues.
2.3. Translation
The goal of assessing the translations’ success in rendering the semantic density of NRSA lies at the interface between Translation Studies and the Pragmatics of Fiction (for an overview of this interface see, e.g., Valdeón, 2017). Regarding Translation Studies, the methodology is set in a DTS framework (Toury, 1995) in the following aspects:
First, it adopts a comparative methodology, by identifying pairs of source text (ST) segments containing NRSA and their target text (TT) correspondences.
Second, it includes two target languages (Spanish and German) in order to control the influence exerted by the target language and literary polysystem on the repertoire available to the translator.
Third, it attempts to find evidence of laws of translation behaviour at work and to identify the conditions increasing their probability of occurrence. For example, the addition of illocutionary features resulting from the choice of hyponyms reported in (Bourne, 2002) might be interpreted as an effect of the law of explicitation, whereby translators tend to produce texts that are in some ways more explicit than their source texts - see (Toury, 1980) and (Séguinot’s, 1988) concept of pragmatic explicitation, among others. For the purposes of this study, pragmatic explicitation occurs when a reporting verb lacking an explicit illocutionary point (i.e., neutral, descriptive, transcript, etc.) is rendered as a reporting expression of the illocutionary class. The converse, i.e., the use of an expression lacking an illocutionary point as the translation of a verb in the illocutionary class, falls in the category of implicitation. The inclusion of two translations of each language is intended to filter language-internal factors (e.g., the existence or absence of certain lexemes), which may constrain translators’ choices.
Other translation regularities might be neutralization (Øverås, 1998), as the tendency to weaken illocutionary strength, particularly when characters speak vehemently. Finally, one must not discard the pull of “expectancy norms”, in particular “quantitative norms of usage” (Chesterman, 1997, p. 83) regarding lexical variety.
Due to the sample size, the present study does not attempt to characterise each individual translator’s style. 4However, it hopes to illustrate how shifts may transform the interpretation of the translated work and the appreciation of evaluative potential of the narrator’s voice through his or her stance. This is in line with Bosseaux’s claim that “features that are inconsistently translated or constantly translated in the same direction will cause shifts in the narrative point of view, focalisation and mind-style in the translations” (Bosseaux, 2007, p. 17). Ultimately, the findings may support discourse models of translated narrative fiction that include the translator and allow for his or her stance to be added to that of the other addressees. In these models, translation shifts may be interpreted as symptoms of translators’ attitudes toward the characters.
3. Methodology
The study requires identification of relevant ST-TT segments instantiating the selected textual phenomenon. First the team5 manually identified instances of NRSA in the source text and extracted their correspondences in the translated versions. The number of instances in the three chapters sampled is 55 (totalling 275 sequences).
There is no automatic procedure to select instances of NRSA, in contrast with what happens with inquit formulae, which can be searched for through typographical conventions (see, e.g., Busse, 2020, pp. 158-159; De Haan, 1997). The range of structures or patterns is illustrated in example 1, with a brief explanatory label.
a. Henry bowed and thanked her. [full verb]
b. but no sooner had he deliberately begun to state his objections [support verb + noun]
c. The scheme was soon repeated to Henry. [verb in the passive voice]
d. Sir Thomas could not give so instantaneous and unqualified a consent. He debated and hesitated; - it was a serious charge; [NRSA without topic]
e. All Huntingdon exclaimed on the greatness of the match [Narrative Voice]
f. Lady Bertram made no opposition [negative NRSA]
g. which could not be satisfied till she had written a long and angry letter to Fanny, to point out the folly of her conduct [Narrative Report of Writing Act]
Narrative Voice (see Semino & Short, 2004) is part of the “minimal forms of speech presentation” involving an indeterminate number of participants collectively sharing a certain illocutionary point. “Negative NRSA are instances of a character refraining from speaking, or not being allowed to speak; they also convey narrator presence” (Toolan, 2001, p. 69). A few instances of Narrator’s Report of Writing Acts (Semino & Short, 2004, p. 48) from letters are counted as well. We excluded reports clearly attributed to the character through Indirect Speech (i.e., a verb introducing a proposition), inquit formulæ (introducing Direct Speech) and Free Indirect Speech (i.e., speech reports compatible with a character’s idiolect).
Instances were subsequently classified according to the Caldas-Coulthard typology (see section 2.1). Three classes were identified: metapropositional illocutionary, neutral structuring and transcript/discourse signalling.
The second step in the analysis was the classification of ST-TT pairs according to the following typology of translation solutions:
a) “NRSA ( same NRSA” (literal translation);
b) “NRSA ( different reporting modes” (Narrative Report of Thought, Indirect Speech, etc.);
c) “NRSA ( different NRSA” (i.e., shifts in the illocutionary features and stance);
d) other - particularly “NRSA ( Ø” and “Ø”(NRSA, i.e., omissions and compensations).
The reason to include category (b) is to contribute to the growing evidence of translation shifts involving forms of speech and thought presentation with a potential to alter the extent of narrator control. Category (c), which is the focus of the study is further divided into: a) addition of illocutionary point (explicitation); b) changes in illocutionary features (illocutionary and stance shifts) and c) elimination of illocutionary point (implicitation).
4. Results
4.1. A classification of NRSA instances in the source text sample
The sample displays remarkable lexical wealth, with 45 lemmas. As presented in Table 1, two thirds of instances convey illocutionary features, while the rest are neutral, structuring and discourse signalling. These results confirm the significance of NRSA in narrator-controlled characterisation in Mansfield Park.
Table 1: A classification of reporting forms according to function
| Type | N | Lexemes |
|---|---|---|
| metapropositional illocutionary | 37 | accept, acknowledge, agree (2), allow (as admit), allow (as permit), assure, bewail, call (sb sth), call on someone to do sth, congratulate, convey (concurrence), debate, dictate, encourage, engage, exclaim on sth, explain, implore, (make or state) objections (2), (speak their) opinion, oppose, (make no) opposition, own, pay (her tribute of admiration), persuade, plead, point out, predict, promote, prove, (bestow) reflections, send (advice and professions), thank, threaten, urge |
| neutral structuring | 16 | address (a question), break news, keep sth to oneself, mention, read to sb, repeat, say (2), speak of sb, talk (1), talk to someone (1), talk of sth (3), treat sth as sth bring a subject forward |
| transcript /discourse signalling | 2 | interrupt, reply (n.) |
4.2. General overview of translation solutions
Before reporting on the findings about illocutionary and stance shifts, this section offers an overview of the translation solutions observed, clarifies potential methodological issues regarding their classification and relates the findings to the topics previously reported in the literature.
The most common translation solution in all four versions was literal translation of the speech reporting form, i.e., an item covering all semantic and pragmatic content of the source text item. In three of the works, (Austen, 1995a, SPT1), (Austen, 1984, GT1) and Austen (1989, GT2), it accounts for about two-thirds of the solutions (64%). In one of the Spanish translations, Austen (1995, SPT2), it is found in 80% of instances. Shifts in speech act features and/ or stance are the second most frequent category of translation solutions, with the three versions clustering at 25-27% and SPT2 a lower 10%.
Shifts in speech and thought presentation modes are similar in all works and so is omission, with one anecdotal instance in each translation. No instances were found of compensation (the insertion of NRSA where it was inexistent in the ST). The figures are presented in Table 2.
Table 2: Translation solutions
| Literal translation | Shifts in the illocutionary features and/or stance | Shifts in speech presentation modes | Other (omission) | |
|---|---|---|---|---|
| SPT1 | 35 [64%] | 15 [27%] | 4 [7%] | 1 [2%] |
| SPT2 | 44 [80%] | 5 [9%] | 5 [9%] | 1 [2%] |
| GT1 | 35 [64%] | 12 [22%] | 7 [13%] | 1 [2%] |
| GT2 | 34 [62%] | 14 [25%] | 6 [11%] | 1 [2%] |
An intra-linguistic comparison reveals that the Spanish translations are more similar to each other than the German ones since there are 19 coincidences of literal translations (e.g., estuvo de acuerdo for “agreed”) and 15 synonym pairs (e.g., SPT1’s cursó amables consejos vs. SPT2’s envió amables consejos for “sent friendly advice”, where cursar is a legal term and enviar is a general language term). 6 Of the shifts, 7 affect the same ST sequence, pointing at some specific asymmetry in the lexical systems of English and Spanish. For example, the ambivalence of bestow in “bestow such very disrespectful reflections” is resolved in SPT1 as the negatively oriented illocutionary verter críticas (“pour criticism”) but as a discourse signalling expression añadir críticas (“add criticism”) in SPT2. Another common feature is that both translators resort to more educated words than those used in the ST (e.g., razonar “to reason” instead of decir for “to say”, conversar “to converse” instead of hablar con for “to talk to someone”). As a general tendency, SPT2 avoids lexical repetition to a larger extent than SPT1, particularly of words occurring in the same clause. For instance, the different senses of talk (talk, talk to and talk of) are translated as five different items (charlar, conversar, comentar, hablar and palabras) in SPT2. In contrast, SPT1 offers two different forms (charlar and hablar).
In the German versions, there are 14 coincidences in literal translation (e.g., erwähnte sie ihn niemals as the translation of “she never mentioned *him*”). In half the cases of literal translation of the speech reporting expression, cotextual material undergoes some kind of translation shift that enhances the presence of the narrator even if his or her stance is not changed. The most significant phenomenon -with six instances in each translation- is a shift from passive to active voice that places the speaker as the sentence’s topic (instead of the illocutionary content of the speech act): “his [Sir Thomas’s] most hearty concurrence was conveyed as soon as possible” is translated as er übermittelte seine freudige Zustimmung sobald wie möglich (“he conveyed his agreement”). Such decisions, which may reveal a quality norm against the overuse of the passive in German, nevertheless alter the topic-comment structure of the sentence and modify the narrator’s indirectness, a resource to create distance.
If we compare all four translations, only 8 instances out of the 55 source text forms involve a literal translation in all four versions, namely predict, threaten, made objections, read, promote, persuade, thank and mention. This is a measure of the extent to which the process of translation is a source of lexical and textual variation.
Although the present study focuses on instances of NRSA, translation solutions are set in larger, text-wide stylistic matrices. Of the two Spanish translations, SPT1 displays more exuberant, oral, intensity-laden stylistic preferences; SPT2 is more restrained, writerly, literal, more compatible with a vision of NRSA where the narrator is a standardization force. As regards the German translations, the study confirms the observation in (Alsina et al., 2017) that GT1 significantly enhances orality in the narrator’s discourse, lending this figure a more conversational tone than in the source text with the insertion of lexical elements from spoken interaction. GT2 is not so profuse in markers of orality (for example, the epistemic dimension of discourse markers is preserved with evaluative adverbs from the written register as opposed to modal particles), although colloquialisms are occasionally found, reinforcing the impression of an omniscient narrator. 7
Regarding shifts in speech presentation modes, the changes observed fall into three types (see Table 3): a) from NRSA to another form not involving speech (Narrator’s Representation of Action and Narrator’s Representation of Thought), b) from NRSA to Indirect Speech and c) from NRSA to Free Indirect Speech.
Table 3: Shifts in speech presentation modes
| NRA/NRT | IS | FIS | N | |
|---|---|---|---|---|
| SPT1 | 1 | 2 | 1 | 4 |
| SPT2 | 3 | 3 | 0 | 6 |
| GT1 | 5 | 2 | 0 | 7 |
| GT2 | 5 | 1 | 0 | 6 |
The first type gives prominence to actions or thoughts rather than speech, hence controlling the amount of speech and changing the way readers access the characters’ consciousness. It is found particularly when English verbs have two senses, one involving speech and one involving (mental) activity. (Semino and Short, 2004, p. 197) point at them as a problematic case in corpus annotation of forms of Speech and Thought reporting. Translators may be forced to disambiguate if the target language lacks comparably ambivalent lexical items. This is illustrated in example 2, where debate is ambivalent between “verbally disagree” and “pondering a question”.
(2)
a. He debated and hesitated (Austen, 1970, p. 4)
b. [GT1] Er widersprach und zögerte. (Austen, 1970/ 1984, p. 8)
He disagreed and hesitated.
c. [GT2] Er überlegte hin und her und zögerte. (Austen, 1970/ 1989, p. 8)
He pondered and hesitated.
In the sampled chapters, instances of this shift and its converse (unambiguous NRSA into ambivalent NRSA) were found more often in German (5 in each translation) than in Spanish (1 and 3, respectively).
The second shift type (from NRSA to IS) seems to be triggered by constraints in the complementation patterns of verbs in each language, as it remains fairly constant in the sampled translations. The contrast between a gerund (English) and a sentential complement (Spanish) are illustrated by example 3 embedded in the reporting of a letter. 8
a. assuring them of her daughter's being a very well-disposed, good-humoured girl
(Austen, 1970, p. 4)
b. [SPT1] asegurándoles que su hija era una chiquilla muy bien dispuesta (Austen, 1970/1995, p. 17)
assuring them that her daughter was a well-disposed girl
This English sentence has been rendered consistently in all four translations as Indirect Speech, i.e., a reporting verb followed by an embedded clause conveying a proposition. In both Spanish translations, the transfer assure - asegurar (a verb requiring a sentential complement) seems to facilitate the shift. In SPT1, the voice syntactically lent to Mrs Price’s proposition is reinforced through items that are plausibly part of her lexicon, such as chiquilla (“little girl”), bearing the diminutive affective morpheme. These instances are beyond the scope of the present paper and therefore no further analysis is conducted on them.
The next section presents the most relevant category of shifts, those affecting information about stance.
5. Shifts in illocutionary features and Engagement value
Qualitatively these shifts are significant because they modify the presentation of the communicative situations involving characters. They are summarized in Table 4 and illustrated with instances from both Spanish and German.
Table 4: Types of shifts in illocutionary features
| Types of Shifts | SPT1 | SPT2 | GT1 | GT2 |
|---|---|---|---|---|
| addition of illocutionary point (explicitation) | 3 | 1 | 3 | 1 |
| changes in illocutionary features | 10 | 2 | 7 | 4 |
| elimination of illocutionary point (implicitation) | 2 | 1 | 2 | 2 |
5.1. Addition of illocutionary point
In this shift a neutral or discourse-signalling speech reporting expression is translated by a lexical item that denotes an illocutionary point, consequently rendering the characters’ intentions more accessible to the reader, as illustrated in example 4. The frequency of this shift is very low.
a. but no sooner had he deliberately begun to state his objections, than Mrs. Norris interrupted him with a reply to them all, whether stated or not. (Austen, 1970/1985, p. 4)
b. apenas había empezado a exponer abiertamente sus reparos, la señora Norris le interrumpió para rebatirlos todos. (Austen, 1970/1995, p. 9)
no sooner had he begun to state openly his objections, than Mrs Norris interrupted him to refute all of them.
The Spanish translation rebatir (“refute”) explicitly conveys the illocutionary point of proving the interlocutor wrong. Hence the narrator is reporting an explicit pronouncement, with Mrs Norris as the source of the absolute truth. This is a more contractive type of engagement than that of reply, which could in principle be a case of entertain, where the speaker is the source of a knowledge piece with high probability of being true. The translator seems to have chosen a verb that is congruent with the hyperbolic and distancing “whether stated or not”, which is part of the narrator-reader level of discourse. The second Spanish translation, not reproduced here, renders it as the cognate réplica (Austen, 1970/1995, p. 49), from replicar “to counter”, closer to that of reply. Translation solutions enhancing Mrs Norris’ contractive stance are consistently found in the Spanish translations not only in NRSA but also in Direct Speech (see Alsina et al., 2017). This may, on the one hand, reflect the attitude of the translators towards this character, and on the other, affect the characterisation cues for the reader.
The tendency toward specification and pragmatic explicitation is attested also in example 5; both Spanish translations specify an illocutionary point and furthermore SPT1 (5b) increases strength.
a. Much was said on his side to induce her to attend the races, [and schemes were made for a large party to them, with all the eagerness of inclination, but it would only do to be talked of.] (Austen, 1970, p. 42)
b. [SPT1] Mucho insistió él para inducirla a que asistiera a las carreras, (Austen, 1970/1995, p. 64)
Much did he insist to induce her to attend the races.
c. [SPT2] Mucho razonó él para inducirla a asistir a las carreras, (Austen, 1970/1995, p. 55)
Much did he reason/argue to induce her to attend the races.
In this passage, Tom, the eldest Bertram son, is planning on being away from Mansfield Park for several weeks just after meeting Mary Crawford. While Tom’s illocutionary point is clearly presented through induce, his dialogic approach is neutral (“saying things”). On the narrator level, however, the neutrality of said is combined with a distancing stance in the semantically vague subject much. In 5b, on the character level, Tom’s approach is direct and contractive through the illocutionary verb insistir “to insist”. The change from passive to active voice and the insertion of Tom as the agent and explicit subject reinforces the directness of the speech act towards Mary. 5c presents, on the character level, a speaker who is dialogically expansive through razonar (“to reason”), suggesting one-to-one interaction. In both cases, the translators’ replacement of the neutral said with more specific verbs closes the interpretation of the situation to reinforce Tom’s agency. These observations confirm the results in (Bourne, 2002). Furthermore, on the narrator level, distance is less evident as the quantifier much (a semantically vague argument of say) is rendered as a booster of the speech act verb.
This section has illustrated the role played by explicitation of illocutionary point in the expression of stance on both levels, the character level and the narrator level; together with changes in cotextual cues, this explicitness can diminish ironic distance.
5.2. Changes in illocutionary features
Changes in the features of the speech acts affect the construal of characters. While they are found in all translations, they are more numerous in SPT1 and GT1. There are shifts in preparatory conditions, in sincerity conditions, and in strength, all of which lead to changes in the dialogic style. This section illustrates such changes with a passage of narrator evaluation and its Spanish and German translations.
Speech acts performed by characters, and their contractiveness or expansiveness, are expected to agree with social status. However, in Mansfield Park hierarchy does not always correlate with stance; Mrs Norris is inferior to Sir Thomas and yet her conversational turns are often reported to be more contractive than his. The narrator hints at this in example 6, a fragment of direct characterisation by the narrator.
a. As far as walking, talking, and contriving reached, she was thoroughly benevolent, and nobody knew better how to dictate liberality to others; (Austen, 1970, p. 6)
b. [SPT1] En cuanto a moverse, charlar y discurrir, era cabalmente caritativa, y nadie sabía mejor que ella cómo enseñar liberalidad a los otros. (Austen, 1970/1995, p. 12)
As far as moving, chatting and talking, she was thoroughly charitable, and no one knew better than she did how to teach liberality to others.
c. [GT1] Solange es ans Planen, Mahnen und Organisieren ging, war sie die Menschenfreundlichkeit selbst, und niemand wusste besser, wie man andere zu Freigebigkeit zwingen konnte; (Austen, 1970/ 1984, p. 9)
When it came to planning, admonishing, and organizing, she was philanthropy itself, and nobody knew better how to force generosity on others.
d. [GT2] Was Spazierengehen, Reden und Pläneschmieden anbelangte, so war sie durchaus entgegenkommend, und niemand wusste besser als sie, wie man andere zur Freigebigkeit anhält (…). (Austen, 1970/ 1989, p. 11)
When it came to walking, talking, and making plans, she was quite accommodating and no one knew better than her how to encourage others to be generous, (…)
This fragment is part of a long paragraph where the narrator evaluates Mrs Morris’s moves to convince Sir Thomas that Fanny will be better accommodated in Mansfield Park than in the parsonage with her (see also Morini, 2009).
At the narrator level, the discourse is monoglossic, as the narrator uses an unmodalised proposition; moreover, there is ironic distance between the actions listed in the topic phrase (making herself busy) and the adjective benevolent, additionally modified by the adverb thoroughly. The verb dictate is open to an ironical interpretation because of the paradoxical relationship between dictate and liberality. As a directive speech act, dictating requires, as a preparatory condition of the speaker, both the will and the capacity to enforce a change of behaviour in the addressee. Mrs Norris may have the former but not the latter, as she is socially inferior (she is Sir Thomas’s poor sister-in-law). Furthermore, there is an incongruence between the verb and its object. Liberality refers to an individual’s attitude, in either of its senses, i.e., tolerance and financial generosity. Attitudes are states of mind that people develop and can hardly be dictated.
At the character level, talk is a neutral verb (no explicit illocutionary point), and it can be considered monoglossic, as it appears that she is talking even if people are not listening. The verb dictate has an explicit illocutionary point and a degree of strength; it can be considered an instance of proclaim, which is contractive.
In 6b enseñar (‘to teach’) excludes the capacity to compel, and hence is more expansive in dialogic terms. It is also less ironical because the collocational clash between dictate and liberality is ‘repaired’.
In the first German translation (6c), on the narrator-reader level, the stance is monoglossic, as it is an assertion. Ironic distance rests on the juxtaposition of the topic phrase, listing three superficial (and superfluous) activities, with the classifying clause war sie die Menschenfreundlichkeit selbst (“philanthropy itself”, with humorous overtones). The contrast between zwingen and Freigebigkeit creates distance to the same degree as the source text because generosity is an attitude that can hardly be enforced.
On the character level, Mahnen (“admonishing”) is an instance of explicitation of an illocutionary point, i.e., type (a) of shifts; rather than a monoglossic talking, the repeated encouragement must be directed at the members of the Bertram household. This shows a heteroglossic stance by Mrs Norris, of the contractive type as she considers herself the source of authority for her admonitions. The second part of the coordination has zwingen, more contractive than dictate (force does not even envisage the existence of alternative opinions, while dictate is still an act of communication). In this translation both monoglossia and contractiveness are globally preserved, while painting a picture of Mrs Norris that is even more authoritarian than in the source text.
In 6d, the stance is monoglossic on the narrator-reader level, as it is an assertion. The adjective entgegenkommend lacks humorous overtones but - combined with the adverb durchaus - it preserves the distancing stance. The verb anhalten has less strength than dictate; distance is not preserved to the same degree as in the source text because generosity is an attitude that can be encouraged and hence the collocational clash is lost.
On the character level, Reden (“talk”) is neutral. The second part of the coordination has anhalten, more expansive than dictate because the speaker signals being more open to alternative positions. This translation presents a picture of Mrs Norris that is less authoritarian than in the source text.
5.3. Elimination of illocutionary point
The frequency of this shift is very low. The loss of illocutionary features has been observed in three types of situations, namely when the speech act becomes neutral or acquires a discourse-signalling role, when the speech report becomes a thought report, and when the NRSA is not reproduced. In example 7, which repeats 4 for convenience, the effects of omission on the stance of the characters as well as on the distance adopted by the narrator on the narrator-reader level are illustrated.
a. - but no sooner had he deliberately begun to state his objections, than Mrs. Norris interrupted him with a reply to them all, whether stated or not. (Austen, 1970, p. 4)
b. [SP1] apenas había empezado a exponer abiertamente sus reparos, la señora Norris le interrumpió para rebatirlos todos. (Austen, 1970/1995, p. 9)
No sooner had he begun to openly state his objections, than Mrs. Norris interrupted him to refute them all.
c. [GT2] als Mrs. Norris ihnen allen, mochte er sie nun dargelegt haben oder nicht, wie folgt entgegentrat: (…). (Austen, 1970/ 1989, p. 8)
as Mrs Norris, all of them, whether he had already presented them or not, countered as follows: (…).
In this fragment, there are four instances of NRSA, two of them attributed to Mrs Norris and two to Sir Thomas; on the character level Mrs Norris shows a monoglossic stance when she interrupts Sir Thomas; her stance is heteroglossic in her reply to his objections, but she might have been presented as expansive or as contractive. However, on the narrator-reader level, the semantically exhaustive adjunct “whether stated or not” is a distancing device that characterises her as contractive, ready to close his argumentative space. Two of the translators shorten the sentence by eliminating an instance of NRSA. 7b omits the hyperbolic adjunct, leading to a loss of narrator distance and hence the cue to her contractiveness; as noted in 5.1, this loss is compensated by the translation of reply as rebatirlos (“refute them”), with the addition of an explicit illocutionary point. 7c omits the fact that Mrs Norris interrupted Sir Thomas, rendering her less monoglossic on the character level. The NRSA in ‘reply’ is rendered as NRS through the insertion of wie folgt (“as follows”).
In example 8 the loss of illocutionary point through lexical choice is illustrated.
a. as she still continued to think Mr. Crawford very plain, in spite of her two cousins having repeatedly proved the contrary, she never mentioned *him*. (Austen, 1970, p. 43)
b. [GT1] aber da sie trotz der wiederholten gegenteiligen Behauptungen ihrer Kusinen immer noch daran festhielt, dass Mr. Crawford ausgesprochen bieder sei, erwähnte sie ihn niemals. (Austen, 1970/ 1984, p. 61)
In this initial appraisal of Mr Crawford by Fanny, the narrator starts with thought report (Fanny’s) that includes speech report by the Bertram sisters. The verb prove denotes that the proposition is not open to discussion; hence, it should be interpreted as an endorsement of the cousins as the external source in their contractive stance that Mrs Crawford is not plain. However, because prove is modified by the frequency adverb repeatedly, a distancing effect is created at the narrator-reader level based on the reader’s previous exposure to the lack of objectivity of the Bertram sisters. 8b, by rendering prove as “observations to the contrary”, loses the illocutionary point and weakens the sisters’ reported contractiveness. The collocational smoothness of wiederholten (“repeated”) and Behauptungen (“observations”) removes the distancing effect created by the narrator.
6. Conclusions
The analysis of the sample shows that the speech reporting from Narrative Report of Speech Acts condenses meaning on two discourse levels (narrator-reader level and character level) that are susceptible of analysis from two perspectives (Speaker and Addressee). The integration of illocutionary features with Engagement allows to account not only for the role of NRSA in characterisation but also for the distancing effects on the narrator-reader level that can be interpreted as irony. Incorporating illocutionary features as information about dialogic stance is a novel methodological contribution to the study of speech reporting forms in general and of NRSA in particular. Approaches to dialogic stance generally have a proposition as the centre (its source and its alignment with the values of both speaker and addressee), leaving the illocutionary dimension aside. The analysis of translations reveals that changes in illocutionary features of the reported speech acts often result in changes in the stance either of the characters or of the narrator. The addresser’s dialogic style is construed by evidence provided by the illocutionary force of his or her speech together with his or her heteroglossia and particularly his or her tendency to contractiveness.
The present analysis is thus a modest contribution to research on one of the least studied forms of speech presentation in Jane Austen’s novels; the classification reveals a clear predominance of illocutionary verbs in NRSA.
Such condensed information is a challenge for translators. From a quantitative standpoint, all translations have at least a 62% rate of success (literal translation), and the Spanish translation by Torres Oliver reaches 80%. Regarding the explicitation of illocutionary features, the sample does not contain enough instances of neutral verbs to test the hypothesis. A few instances point to strategies like disambiguation (introducing a clear illocutionary point), avoiding repetition and ensuring collocational congruence.
In order to grant such observations, the status of consistent changes at the global level of character portrayal a larger sample would have to be analysed; they do, however, prove changes in the tone of certain episodes associated to dialogic styles of their participants, with potential mismatches between their stance and their power status.
The paper is also a contribution to the study of the discourse of narrative fiction, as it proves the existence of lexical manifestations of the various levels of addresser-addressee. Translators will benefit from an awareness of this multi-level structure and of the semantic condensation expressed by an apparently simple form of speech presentation such as NRSA. The integration of these varied perspectives has practical applications in translation pedagogy, for example in literary translation courses, and translation quality assessment, by providing explicit parameters to assess the stylistic adequacy of lexical choice.
