Introduction

When communicating, either in spoken or written mode, people use discoursal elements by which they can construct, maintain, and direct their interaction(s) to their receptive audience. Considering this, it said that (see for example, Vande Kopple 2002; Hyland 2005, 2019; Herriman 2014) texts are usually produced at two levels of meaning; namely propositional or content level and interactional level. To elaborate on this dichotomous nature of texts, Herriman (2014) states that

Texts may be seen as consisting of different levels of meaning, a propositional content level, which refers to actions, events, states of affairs or objects in the world portrayed by the text, and a writer-reader level, where the writers interact with their readers, explicitly guiding them through its structure and organization, commenting on the writing process itself or expressing their opinions and beliefs concerning its content (p. 1).

The level on which the communication between the producer of the message and the receiver of it is established is called interaction, which is done, constructed and directed through some very particular features. In academic literature, these particular elements are called metadiscourse features (Hyland 2017), discourse markers (Hyland and Tse 2004), or metatalk (Vande Kopple 2012). The main specification of these discoursal features is that they do not add anything to the propositional meaning(s); instead, they are used to establish the writer-reader or speaker-audience interaction.

The term metadiscourse was first coined by Zelig Harris in 1959 and after a short-period gap, it received further attention, commenced by Williams (1981) and resumed by others, particularly Vande Kopple (1985) and Crismore (1989) to refer to “a self-reflective linguistic expression referring to the evolving text, to the writer, and to the imagined readers of that text” (Hyland 2004, p. 133). In the same vein, Vande Kopple (2012) defined metadiscourse features as “elements of texts that convey meanings other than those that are primarily referential” (p.37). In another definition, Schiffrin (1987), described metadiscourse features as “sequentially dependent elements which bracket units of talk” (p.31).

Considering writer-reader interaction in communication, languages are classified into two main categories; namely, as writer-oriented and reader-oriented (Hinds 1987; Qi and Liu 2007). One way of determining whether a language is writer-oriented or reader-oriented is through looking at metadiscourse features (Zarei and Mansoori 2011; Shokouhi and Talati Baghsiahi 2009). As a matter of fact, in a writer oriented language, in the course of the communication between the writer and the reader, it is the responsibility of the writer to make the communication as effective and fathomable as possible to the readership (Herriman 2014), whereas in a reader-oriented language, it is the responsibility of the reader to make an effort to unearth the meaning and/or the essence of the speaker and/or writer (McCarthy 2005). Regarding this dichotomy, both English and Persian languages are classified as writer-oriented languages (Zarei and Mansoori 2011). To put it differently, in English and Persian languages, it is the responsibility of writers to produce their texts in such a way that it can be understandable for their readership (Zarei and Mansoori 2011).

Metadiscourse has been studied increasingly by different scholars and is not a newly described entity; adopting various approaches in language studies (see for example Hyland 1999; Lin 2005; Farrokhi and Ashrafi 2009; Crismore and Abdollahzadeh 2010; Abdi 2011; Akbas 2012; Letsoela 2014; Alyousef 2015; Ma and Wang 2015; Sešek 2016; Ghahremani Mina and Biria 2017; Ramoroka 2017; Jalilifar et al. 2018; Vasheghani Farahani 2018; Mahmood et al. 2018; Alotaibi 2018; Vasheghani Farahani and Ahmed Ibrahim 2018; Molino 2018; Akbas and Hardman 2018).

Moreover, in terms of English and Persian languages, an array of previous studies demonstrated that Iranian writers usually use fewer metadiscourse features when compared to those of the English native speakers (see for example Khajavy et al. 2012; Vaez Dalili and Vahid Dastjerdi 2013; Yazdani et al. 2014; Kuhi and Mojood 2014; Ghadyani and Tahririan 2015; Vasheghani Farahani and Sabetifard 2017; Vasheghani Farahani 2017; Tadayyon and Vasheghani Farahani 2017). According to the results found in these studies, one very salient reason is the socio-cultural settings in which the communication took place as well as the influence of the first language of the authors (Persian) on their second language writing.

Unlike the traditional and commonplace way of extracting and counting metadiscourse features and/ or any other language elements by hand which is time-consuming and subject to human errors, one acceptable and trustworthy way of extracting and analyzing metadiscourse features in different texts, genres, and/or languages, in large quantity, is to use corpus and electronic software (Heng and Tan 2010). From among different types of corpora, Parallel corpora are among the best tools for analyzing and comparing specific language entities like metadiscourse markers in two or more languages (Zanettin et al. 2003; Anderman and Rogers 2008; Candel-Mora and Vargas-Sierra 2013). With regard to this study, parallel corpora are found to be useful tools as they permit us to search for the nuances in a cross-linguistic way.

Since their advent, parallel corpora have been widely used in translation; however, a scan of the related literature shows that most of these studies were based on written discourse not spoken. This can be due to the fact that creating a parallel corpus of spoken data is inherently an arduous and convoluted kind of task as spoken discourse must first be transcribed. Considering the above-mentioned issues, this study was an effort to fill out this gap. Therefore, with regard to the writer-oriented and reader-oriented language dichotomy and the writer-reader interaction in political spoken discourse, this study sought to address two research questions to explore: i. the distributional behavior of metadiscourse features in TED spoken political talks in English and their Persian translation and ii. to see if the level of interaction in the source text and the target text changed or remained stable? The hypothesis put forward in this study was that due to the fact that both English and Persian languages belong to the writer-oriented languages, there should not be significant differences between the distributional pattern of metadiscourse features in English and Persian languages and that the speaker-audience interaction remains intact in translation from English into Persian.

Review of Literature

In terms of metadiscourse and translation (either written or spoken mode) there are reportedly some semi-related works (see for example e.g., Williams 2010; Gholami et al. 2014; Davtalab et al. 2012; Herriman 2014; Kafipour 2016; Mardani 2017; Dupont and Zufferey 2017; Vasheghani Farahani and Dastjerdi 2019, Vasheghani Farahani and Kazemi 2020). The results of these studies mostly demonstrated that, in terms of translation, due to the omission of metadiscourse features, there were, reportedly, fewer metadiscourse features in the target language as compared to the original language and that the interaction between the writer and the reader changed in translation.

Despite the paucity of research, in a very related study to this research in hand, Crible et al. (2019) analyzed the function and translation of metadiscourse features in TED talks to determine the level of underspecification as a phenomenon in linguistics defined as the process in which an underspecified meaning is assigned to an ambiguous word or expression to prevent unnecessary unambiguous steps (Egg 2010). For this purpose, they created a multilingual parallel corpus of five languages including Czech, French, Hungarian, and Lithuanian and English. Being limited to only three metadiscourse features including “but”,” and”, “so”, they showed that there were three kinds of underspecification as functional mismatch, omission, and weaker translation. Moreover, the results of their research revealed that not all of the metadiscourse features were affected by the underspecification.

In another study that is similar to the current one, Hoek et al. (2017) investigated the type of metadiscourse features relations which were omitted in the translation. For this reason, they created a multilingual corpus of English parliamentary debates translated into Dutch, German, French, and Spanish. Their investigation proved this hypothesis that some relations in metadiscourse features had more tendency towards omissions such as speech-act relations and positive causal relations.

In the same vein, Furkó (2020) did research on issues of (under)specification in the translation of reformulation markers in scripted discourse. For doing so, he limited his study on two metadiscourse markers including “I mean” and “actually” which were translated from English into Hungarian. The analysis of the parallel corpus showed that there was some sort of underspecification in translating these two metadiscourse features from English into Hungarian; meaning that not all of the discourse markers were translated into Hungarian and that there were functional mismatches in target language.

Methodology

Data Gathering Regime and Corpus Creation

As this study was corpus-based research, it was necessary to have a balanced and representative corpus to meet the requirements of the research (see Anthony 2009). There are a huge number of parallel corpora that are commercially available; however, when it comes to English-Persian language pair, there seems to be paucity of corpus data as the Persian language is regarded as a less researched language in a global scale. Moreover, most of the corpora are extracted from written sources; making the creation of this specialized, parallel, uni-directional, small-scale, Do it Yourself (DIY) corpora as an indispensable part of this research. For this reason, 30 various TED talks were randomly selected. The TED talks were selected from 2010 to 2018 time span; making this fine-grained corpus a more synchronic one. They were all translated and subtitled by Persian native speakers/ translators and were commercially available. It is noteworthy to mention that there was no trace of manipulation of the translations from the researcher(s); meaning that the translations were uploaded into the software and analyzed in the way they were translated by translators. The TED talks were all in the domain of politics; including a wide range of various sub-corpora and/ or topics from U.S foreign policy, US elections to democracy, and Brexit (UK exit from the EU Union). The TED talks were all presented by male and female native speakers of English in 10 to 20-min presentations. The 30 TED talks were all merged into one unified corpus in order to create a synchronic, unannotated, specialized, uni-directional parallel, representative and balanced corpus of political texts in English and Persian languages. Once merged, the texts were aligned at the sentence and /or paragraph levels in an Excel file. Although there are some alignment tools, their quality, at least for English-Persian language pair, is under question. For this reason, the alignment process was done manually. This suitable testing ground English corpus consisted of 25390 tokens, 21,665 words, 2979 lempose, 2747 lemma, and 1460 sentences. As far as the Persian corpus was concerned, there were 24,244 tokens, 21,164 words, and 1538 sentences. For analyzing the corpus, Sketch Engine corpus platform was used to analyze sentences as it allows the creation and analysis of parallel corpora (Heng and Tan 2010) (Fig. 1).

Fig. 1
figure 1

A screenshot of the parallel corpus

Metadiscourse Category

For analyzing metadiscourse features, the model developed by Hyland (2005) was used. His classification is composed of two main categories each of which has five subcategories. The main categories are interactive and interactional. In the interactive classification, the writer will try to shape the text in such a way that it could meet the needs of the readership (Hyland 2005). The interactive category is composed of five sub-categories including transitions (devices to improve the connections between sentences and paragraphs.), frame markers (devices used to refer to sequences and acts), Endophoric markers (devices to refer to information in other sections of the text), evidentials (devices to refer to the information to other sources) and code glosses (devices to elaborate propositional meaning).

On the other hand, the interactional category of metadiscourse features is used to make the message of the writer more explicit, which is done by allowing the receiver to respond to the unfolding text. In other words, this category contains features that involve readers and open opportunities for them to contribute to the discourse by alerting them to the author's perspective. This category is composed of five sub-categories as hedges (devices used to show uncertainty), boosters (devices used to show certainty), Attitude markers (devices used to show writer's affective attitude), Self-mentions (devices to reveal the writer's presence in the text) and Engagement markers (devices used to indicate the interaction of the readership) (Table 1).

Table 1 The category of Hyland’s metadiscourse features

Procedure

For putting this study into practice, once the corpus was created and uploaded into Sketch Engine, the whole English corpus was read manually to detect any instances of metadiscourse features. Therefore, this process was inherently a bottom-up process as there were no preselected elements worth analyzing. Once the tokens of metadiscourse features were identified in the English corpus, their frequency (types) was detected via concordance lines in the Sketch Engine. The manual reading of the corpus showed that there were 39 instances (tokens) of interactive metadiscourse features and 60 interactional ones (regardless of their frequency). The metadiscourse features were then categorized based on Hyland’s classification to create a quantitative database of metadiscourse features in the English corpus. Once the number of metadiscourse features was quantified in the English corpus, their Persian translation was also detected and quantified for the sake of comparison. After that, the corpus was read carefully to analyze the texts qualitatively and detect any traces of changes in translation.

Data Analysis

In order to run the statistical analysis, the frequency test together with the Chi-square test was used to show the level of significance. For the statistical analysis, SPSS was applied. First, the interactive category was analyzed, then the interactional. The results are shown below.

Interactive Metadiscourse Features

As the data in Table 2 demonstrate, the number of interactive metadiscourse features in the English corpus was 395 instances out of which transition with 246 items (62.3%) were the most frequent instances of metadiscourse features followed by code glosses and frame markers as the second and third most applied interactive metadiscourse features with 82 (20.8%) and 63 (15.9%); respectively. Apart from the distributional pattern, to see the percentage of transfer and omission of metadiscourse, the frequency of metadiscourse features was calculated. The results are shown in Table 3.

Table 2 Frequencies and percentages of the features in both languages
Table 3 Frequencies and percentages of the interactive features translated or not translated to Persian

As the data in Table 3 reveal, in the English corpus, were 246 items of transitions out of which 232 items were translated/ transferred, and the rest were omitted. When it comes to frame markers, from 63 items, 57 items were transferred/translated which equals 90% of the frame markers. Endophoric markers and evidentials were 4 items in the source text and fully translated/ transferred into the Persian language. From 82 items of code glosses, 75 items (91%) were transferred/ translated. In general, for all categories, 93% of the interactive metadiscourse features were translated, while 7% were not translated. However, to reach a better understanding of the translation/ omission rate, the data were tabulated in a bar chart which is shown in the figure below.

Figure 2 shows the percentage of interactive metadiscourse features translated and not translated from English into Persian. As can be seen, 94 % of transitions were translated and 6% were not. For frame makers, 90% were translated and 10% were omitted. For endophoric markers and evidentials, 100 % were translated into Persian. As for the code glosses, 91% were translated and 9% were omitted. However, for understanding the level of significance, a Chi-square test was performed. The results are shown in Table 4.

Fig. 2
figure 2

Percentages of the interactive features translated or not translated to Persian.

Table 4 Results of the Chi-square test

Table 4 demonstrates the result of the Chi-square test. The result of the test shows that there was not a significant difference between the two languages because the p-value was greater than .05 (X2 = .09, p = .999). Chi-square goodness of fit is run to analyze each feature separately.

Table 5 shows the result of the chi-square goodness of fit test for the interactive category of metadiscourse features. The result of the test reveal that there was no significant difference the use of transitions (X2 = .41, p = .522), frame makers (X2 = .30, p = .584), endophoric markers (X2 = .00, p = 1.000), evidentials (X2 = .00, p = 1.000), and code glasses (X2 = .31, p = .576) in English and Persian because all p values were greater than .05.

Table 5 Results of the Chi-Square goodness of fit test

Interactional Metadiscourse Features

Table 6 demonstrates the frequency and distributional pattern of interactional metadiscourse features in the English corpus and its Persian translation. As can be seen, hedges were found 82 times in the English corpus; whereas, in the Persian translation, there were only 62 items. When it comes to boosters in the English corpus, there were 304 items; whereas in the Persian translation there were 283 items. 17 instances of attitude markers were found in the English corpus; however, in the Persian translation, there were only 14 items. For the self-mentions, in the English corpus, there were 90 items; whereas in the Persian, there were 86 items translated and/ or transferred. For the engagement markers, from 157 items, 100 % translated and transferred into the Persian language. Beside this table, the percentage and frequency of interactional metadiscourse features were analyzed. The results are shown in Table 5.

Table 6 Frequencies and percentages of the metadiscourse features in both languages

Table 7 demonstrates the distributional pattern and the percentage of translation and omission of the interactional metadiscourse features in English and its Persian translation. As can be seen, 76% of the hedges were translated and 24% were omitted. For boosters, 93% were translated and 7% were left untranslated. In attitude markers, 82% were translated/transferred; whereas, 18% were omitted. For the self-mentions, 96% were translated and the rest (4%) were omitted. Engagement markers were fully translated. Overall, 93% of the interactional metadiscourse features were translated and /or transferred and 7% were omitted. Like the interactive category, for reaching a more robust understanding of translation and / or transfer of interactional metadiscourse features, their percentage was tabulated in a bar chart below.

Table 7 Frequencies and percentages of the interactional features translated or not translated to Persian

Table 8 shows the results of the Chi-square test. The result of the test shows that there was not a significant difference between the two languages because the p-value was greater than .05 (X2 = 2.07, p = .722). Chi-square goodness of fit is run to analyze each feature separately (Table 9).

Table 8 Results of the Chi-Square test
Table 9 Results of the Chi-Square goodness of fit test

The results of the test reveal that there was not significant difference the use of hedges (X2 = 2.78, p = .096), boosters (X2 = .75, p = .386), attitude markers (X2 = .29, p = .590), self-mentions (X2 = .09, p = .763), and engagement markers (X2 = .00, p = 1.000) in English and Persian because all p values were greater than .05.

To further analyze the English corpus with its Persian translation, the type-token ratio was calculated. To put it simply, token refers to the total number of words, and type refers to the total number of unique words (Baker 2006). Type token ratio refers to

The average number of tokens per type… A corpus or file with a low type/token ratio will contain a great deal of repetition - the same words occurring again and again, whereas a high type/token ratio suggests that a more diverse form of language is being employed (Baker 2006, p. 52)

Table 10 demonstrates the type-token ratio of the English corpus. As can be seen, the rate of the type-token is 85, 32 which is rather high. It means that the English corpus had rather a highly diverse form of language in use.

Table 10 Type- token ratio of the English corpus

Table 11 shows the type-token ratio of the Persian corpus. As can be seen, the rate of the type-token ratio is 87, 29. This ratio is more than that of the English corpus which means that the Persian language had a more diverse range of vocabulary and less repetition compared to English corpus.

Table 11 Type- token ratio of the Persian corpus

Results

Considering the notion of writer-reader interaction which is established through the notion of metadiscourse features and concerning writer-oriented and reader-oriented language dichotomies, this research was an attempt to unearth the distributional pattern of metadiscourse features in TED political discourse as well as determining the preservation or change of the interaction in translation from English into Persian.

Overall, regardless of their frequency, the English corpus contained 39 instances of interactive and 60 instances of interactional metadiscourse features. Correspondingly, the corpus was found to be more interactional oriented (650 items) rather than interactive oriented (395 items). Moreover, the English corpus had more metadiscourse features as compared to the Persian language in both interactive and interactional categories (1045 vs. 970). The more reliance of the English native speakers on metadiscourse features can reveal that authors were more engaged in interacting with the readers and representing themselves as competent speakers in their fields of study. In addition, the more prevalence of the metadiscourse features in English talks may indicate the fact that the speakers were more concerned not only with the propositional meaning; but also with the discoursal level as the discoursal level (non-propositional meaning) is constructed through metadiscourse features.

Moreover, the differences between the type-token ratio showed (see Tables 10 And 11) that there was a difference between English and Persian language in terms of vocabulary usage and density. Indeed, the higher type-token ratio of the Persian corpus (translation) can indicate that the translated corpus in the Persian language had a more diversified range of vocabularies as compared to that of the English original corpus. This is in line with the less number of metadiscourse features found in the Persian language when compared to English. In other words, it is inferred that as there was a lower range of repetitions of metadiscourse features in the Persian language, there was a higher range of metadiscourse features in the English language.

Considering the fact that metadiscourse features will not change the meaning/proposition structure of the sentence; instead, are used, basically, for shaping the structure of the discourse as well as guiding the receiver of it in a clear way (Furkó 2020), the less deployment of interactive metadiscourse features by speakers can suggest that in English talks, more efforts were made to help audience to understand the structure and rhetoric of the message.

The employment of transitions, frame markers, endophoric markers, evidentials, and code glosses have very certain implications. As a matter of fact, the usage of transitions showed that those speakers were concerned, mostly, with producing texts which were coherent, and which could transfer the message in a grammatically precise and understandable way. Besides, this inclination of authors to use transitions apparently can show that they helped the audience to unfold and interpret the pragmatic connections of various steps in the argumentations as they can be reached, wholly or partially, by transitions.

The reliance of speakers on code glosses as the second most widespread interactive metadiscourse features indicated that they made efforts, but less than using transitions and endophoric markers, to provide extra information for the fluent understanding of the readers. Another function of code glosses is to elaborate on the meaning which is thought to be challenging to grasp for the audience. By using this device, speakers, suggestively, endeavored to clarify this problematic notion by giving complementary information to the audience.

The usage of frame markers in the English corpus revealed that the speakers tried to structure the discourse to facilitate the understanding process of the reader as well as different sequencing parts of the text to internally establish an argumentation. This exploitation of frame markers in interval sequences in language could suggest that speakers knew how to sequence materials in their texts to be able to construct their argumentation.

Like the interactive metadiscourse features, there were fewer instances of interactional metadiscourse features in the Persian translation of the English corpus. The less deployment of interactional metadiscourse features revealed the fact that the speakers of the English talks considered the possible audience by making their attitudes towards the proposition as these roles are played by these features. Notwithstanding, the less reliance of the Persian translations on the interactional metadiscourse features could reveal the fact that this function was not fully met in the translation.

In the interactional category of metadiscourse features, boosters were the most applied ones. Boosters are used by the authors or speakers to close down any alternative proposition. In other words, these features signal the certainty of the author or speaker about his proposition. By applying these features, the speakers showed that they did not recognize any conflicting or opposite ideas and that they had the final say. In this regard, it can be said that in the English corpus, boosters were used as the most frequent instances of interactional metadiscourse features to show that the authors had no willingness to recognize an alternative proposition and that they were mostly consistent on their propositions.

Engagement markers were the second most used interactional metadiscourse features in the English corpus. Usually, engagement markers are exploited to show the engagement of the leadership in the discourse. To put it differently, by using these features, the speakers showed their interest in including and in participating in the interaction with the audience. This is done by highlighting or downplaying the presence of the readership collaboratively.

Self-mentions were regarded as the third used instance of interactional metadiscourse features. Usually, self-mentions were used by the speakers to show their distinctive personalities in the discourse. In other words, self-mentions were crucial in that they helped TED speakers to promote and establish their own ideas/thought while referring to others'. It can be said that authors in the corpus felt the necessity to develop their ideas and identity while quoting and citing other ideas as this could be reached by self-mentions.

Hedges were the fourth used interactional metadiscourse features in the English corpus. Unlike boosters, hedges signaled the presence of the opposite voices in the corpus. In other words, hedges were used to show that the author/ writer had no commitment towards the message as he allowed the alternative voice(s) to be interjected in the TED talks. By using these features, the speakers showed that their statements were plausible and that they were not 100 % certain (relative uncertainty) in their statements. In addition, by using these features, speakers showed that their statements were mostly based on the interpretation, not certain facts as they allowed the addition of extra propositions and / or claims in their speech.

Although there was statistically no significant difference between the distributional pattern of metadiscourse features in English corpus and its Persian translation and Persian language (translation) followed the same ranking pattern of the English corpus, the qualitative analysis of the corpus showed that there found to be four kinds of changes in translating metadiscourse features from English into Persian as implicit /explicit changes, disinformation changes, (de)emphasis change and (in) visibility changes. These categories are used in this study as tools for qualitative analysis (Herriman 2014). The first change can be (im)explicit change which can be traced in the deletion or preservation of transitions, endophoric markers, evidentials, and endophoric markers.


(Im)Explicit Change

One possible change in the translation from English (source text) into Persian (target text) is the explicit/ implicit change. To elaborate, the preservation of transition markers, evidentials, endophoric markers, and frame markers can ensure the level of explicitness in the translation. Conversely, the omission of these features can ensure the implicitness of the information in translation. To elaborate, some examples are given below:


Match


1: Although the nation has become stronger and more confident since 2000, it is not a rising power given the growing international competition and domestic constraints on its development.

figure a

Back Translation


Although the nation has become stronger and more confident since 2000, international competition and domestic constraints on development will continue in a more quiet context.

In this example, there is a one to one match/equivalence in the transition marker “although“ and its Persian translation. In this regard, it can be said that the level of explicitness remained in translation as the transition marker was translated. In other words, it can be said that as far as these elements were concerned, the level of text explicitness did not change in translation; despite the deletion of some transitions in translation.


2: All activities should follow strict and transparent project management rules and regular programme and project evaluations should be held periodically.

figure b

Back Translation


All activities should follow strict and transparent project management rules and regular programme and project evaluations that should be held periodically.

In this example, there is an example of endophoric marker “ follow” which is transferred and translated into Persian. As such, the level of explicitness remained intact in translation into Persian.


3: They said that the Assembly had endorsed the Plan under circumstances unworthy of the United Nations and that the Arabs of Palestine would oppose any scheme that provided for the is section, segregation, or partition of their country, or which gave special and preferential rights and status to a minority.

figure c

Back Translation


They said that the Assembly had endorsed the Plan under circumstances unworthy of the United Nations and that the Arabs of Palestine would oppose any scheme that provided for the is section, segregation or partition of their country, or which gave special and preferential rights and status to a minority.

In this example, there is a one to one match equivalence of the evidential marker “they said that” in English and in the Persian translation. With the transfer of this evidential metadiscourse feature, the level of explicitness remained in place in translation.


Omission


1: In this Part, however , I discuss evidence from a variety of sources suggesting that parties can and do bargain around liability as well as property rules in IP.

figure d

Back Translation:


In this section, I will discuss some evidence from different sources which suggest that parties can bargain about the liability and property rules in IP.

As can be seen in this example, there is no one to one equivalence for the transition marker “however” in the Persian translation. Therefore, it can be said that by the omission of “transition” there found to be some trace of implicitness in translation as the translator(s) found it not necessary to transfer.


2: I particularly and personally was accused of siding with, for instance , the citizens of Sarajevo -- "siding with the Muslims," because they were the minority who were being attacked by Christians on the Serb side in this area.

figure e

Back Translation


I particularly and personally was accused of siding with the citizens of Sarajevo - - "siding with the Muslims," because they were the minority who were being attacked by Christians on the Serb side in this area.

As can be seen in this example, the code gloss “ for example “ was not translated into Persian. This omission by the translator (s) could reveal the fact that they did not consider it as a necessary piece of information; decreasing the level of explicitness and increasing the level of implicitness.


(Dis)information changes


Another type of modification in translating metadiscourse features from English into Persian is disinformation change. Indeed, this kind of change is done in situations when the number of code glosses differs in translation. The main function of code glosses is to add new information to the readership as the translator may think that they need more information for the ease of interpretation. Therefore, the reverse is done when the translator thinks that the possible reader may not that information.


Match


1: But also understand their mistakes, because, as women, in particular, we're taught that if something doesn't work out, it's probably our fault.

figure f

Back Translation


But to understand their mistakes as, women in particular, have learned that if something will not work properly, it is properly our mistake.

As can be seen, there is a one to one correlation/ equivalence in English booster and its Persian translation; therefore, the level of information remained in place in source corpus and target corpus.


2: So for instance, London and the Southeast have the highest numbers of immigrants, and they are also by far the most tolerant areas.

figure g

Back Translation


So for instance, London and the Southeast have the highest numbers of immigrants, and they are also by far the most tolerant areas.

As can be seen in the above example, there is a one to one equivalence between the code gloss “So for instance “ in English and its Persian translation and the level of information remained place in English and Persian.


Omission


1: Last month, for example, a leading relief NGO responsible for running a camp of 90,000 displaced people was forced to pack its bags.

figure h

Back Translation


Last month a leading relief NGO responsible for running a camp of 90,000 displaced people was forced to pack its bags.

As can be seen in this example, there is no one to one equivalence for “ for example” in English corpus and its Persian translation. As a result, it can be said that there was a trace of disinformation in the Persian language.


(De)emphasis Change

Another type of change and/ or modification in translating metadiscourse features are emphasis changes. Emphasis change is done through the insertion of boosters. When a booster is added or deleted in a proposition, the force of the proposition can be increased or decreased; depending on the choice of the translator.


Match


1: He says it is " vitally important that the international community not allow the differences of the past months to persist, and that it finds unity of purpose around a common security agenda".

figure i

Back Translation


He says it is " vitally important that the international community not allow the differences of the past months to persist, and that it finds unity of purpose around a common security agenda".

As can be seen in this example, there is a one to one translation between the source text booster and its Persian translation. In this regard, it can be said that the force of the source text proposition remained in place in Persian translation.


2: There is, of course , a third choice, perhaps the most likely one, which is that we do neither of those things, and in four years time you invite me back, and I will give this speech yet again.

figure j

Back Translation


There is, of course , a third choice, perhaps the most likely one, which is that we do neither of those things, and in four years time you invite me back, and I will give this speech yet again.

As can be seen in the above example, there is a one to one equivalence in the source text and that of the target text. As a result, it can be said that the force of the proposition remained intact in translation.


Omission


Over the past few years, there's been a fascinating convergence of findings in several different sciences, in psychology and anthropology and neuroscience and evolutionary biology, and they all tell us something pretty amazing.

figure k

Back Translation


Over the past few years, there's been a fascinating convergence of findings in several different sciences, in psychology and anthropology and neuroscience and evolutionary biology, and they all tell us something amazing.

As can be seen in the above example, the booster “ pretty“ was not translated into Persian. As a result the force of the sentence in Persian was declined.


Omission


Well, you talk about the world in crisis, which is absolutely true, and those of us who spend our whole lives immersed in this crisis.

figure l

Back Translation


Well, you talk about the world in crisis, which is true, and those of us who spend our whole lives immersed in this crisis.

As can be seen in the above example, there is no one to one equivalence between the booster “ absolutely” and its Persian translation. As a result, the force of the proposition decreased in Persian.


(in)Visibility Change

(In) visibility change is done when self-mentions and attitude markers are either inserted, translated or omitted in translation. By these features translators can show or conceal their presence in translation; depending on their choices.


Match


1: So that is where, for me, I understood that objectivity means giving all sides an equal hearing and talking to all sides.

figure m

Back Translation


So that is where, for me, I understood that objectivity means giving all sides an equal hearing and talking to all sides.

As can be seen in the above example, there was a one to one equivalence between the source text self-mention “I understood that “ text and the target text. As a result, translator demonstrated authors’ presence in the target text.

As can be seen in the above example, the translator did not translate the self-mention “ I understand that“; reducing the presence of author into Persian.


Omission


1: But no matter who it was, I think, finances are often a reason we don't let ourselves dream.

figure n

Back Translation

But no matter who it was, finances are often a reason we don't let ourselves dream.

As can be seen, there is no one to one equivalence for the self-mention “ I think” in the Persian translation.

Concluding Remarks

Overall, the quantitative and qualitative analyses of the corpus showed that there was statistically no significant difference between the number of metadiscourse features in the English corpus and its translation into Persian. Therefore, it can be said that the null hypothesis of this study was confirmed; meaning that the writer-reader interaction in translation from English into Persian in political discourse was, at least to a great extent, remained constant. This can be partial because, statistically, there was no significant difference between English and Persian languages in terms of metadiscourse features as depicted by Tables 2 and 6. This statistically equal number of metadiscourse features in both languages can add support to the point that translators were mostly aware of the role of these features in translation; therefore, translating the talks in such a way that the significance of non-propositional elements would not be underestimated for the receptive Persian audience. Another reason why the null is confirmed can be attained to the writer-oriented nature of English and Persian languages. Being writer oriented entails a clear and explicit transfer of meaning to the audience in terms of propositional and non-propositional meaning. As a result, translators, being aware of this explicitness of meaning transfer, tried to transfer and translate the metadiscourse features into Persian to preserve the significance of non-propositional meaning.

The findings of this research are hoped to have useful and practical implications for researchers in the field of corpus-based translation studies. The researchers to be will find the methodology of this research useful for building parallel corpora and corpus interpretation as well as corpus investigation. In addition, those interested in contrastive analysis and corpus linguistics can benefit from the findings and the results of this study.

This research was limited to and focused on spoken discourse as a distinct mode. It might be a good idea if the written mode is also analyzed as these two modes differ from each other. In addition, this research was based on a corpus of political discourse as a distinct genre. The same research can be run in the various genre with the purpose of comparing the usage and distribution of metadiscourse features as well as the interaction between the authors and readers.