Volume 13, Issue 1 (2022)
LRR 2022, 13(1): 701-745 |
Back to browse issues page
Download citation:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:
Taghavi M, Hashemi M. A Critical Corpus-based Study of Translation Universals: A Comparison of Translational and Original Persian. LRR 2022; 13 (1) :701-745
URL: http://lrr.modares.ac.ir/article-14-44062-en.html
URL: http://lrr.modares.ac.ir/article-14-44062-en.html
1- .D. in Translation Studies, English Department, Ferdowsi University of Mashhad, Mashhad, Iran
2- Professor of Applied Linguistics, English Department, Ferdowsi University of Mashhad, Mashhad, Iran ,hashemi@um.ac.ir
2- Professor of Applied Linguistics, English Department, Ferdowsi University of Mashhad, Mashhad, Iran ,
Abstract: (3125 Views)
One of the crucial topics discussed in descriptive translation studies is that of Translation Universals (TUs), which addresses typical, salient features of translational language that make it distinguished from other linguistic variants. Taking into consideration the differences between languages, the key question here is whether the purported universal features (mainly articulated based on examining European languages) exist in non-European, less- or uninvestigated languages. Employing Chesterman’s categorization of universals into ‘S-universals’ and ‘T-universals’, the present study aimed at examining the latter, less investigated group of universals. A comparable corpus was made of original and translated Persian expository texts to investigate two T-universals, namely simplification and explicitation. In the light of linguistic features of translational Persian obtained, the present study challenges the purported universals as none of the extracted features were in line with the previous studies’ prepositions.
1. Introduction
One of the most significant topics whithin Descriptive Translation Studies (DTS) is Translation Universals (TUs), first clearly articulated by Mona Baker in her work (1993). TU hypotheses are concerned with typical linguistic features that makes translational language different from other linguistic variants. According to Hansen and Teich (2001), “it is commonly assumed in translation studies that translations are specific kinds of texts that are different not only from their original source language (SL) texts, but also from comparable original texts in the same language as the target language (TL)” (p. 44). Such claims can be examined either manually or by means of corpus-based analytical tools. Corpora have been a reliable popular tool among researchers since the convergence of corpus-based empirical methodology and linguistic studies, including Translation Studies, during the 1990s.
Over the last three decades, many studies have been conducted on theories of TUs and evidence of specific linguistic features of translational language has been provided, but almost all of these studies have been on Western languages, especially English. Chesterman (2004) divides the TUs into "S-universals" and "T-universals." The first category refers to "universal differences between translations and their source texts" (ibid, p. 39) and the second category refers to the differences in the linguistic features of translations (target texts) as compared to non-translated, native TL texts. Although some TUs may usually be investigated within one category, some can be examined from the perspective of both groups.
Some interpret the so-called TUs as an inseparable part of any translational language and are in line with the theories presented in the literature of translation studies. It should be noted, however, that if a linguistic feature is to be considered a "universal," it should be found in translations into all languages, but, in fact, almost all the literature on TUs is devoted to research on Western languages, especially English. Only a few examples can be found (e.g. Xiao & Hu, 2015) that have studied universals in non-European languages. In addition, research on S-universals outweighs work on T-universals. Hence, such claims as the existence of ‘universal’ features, in the strict sense of the word ‘universal’, is highly debatable, unless they are scrutinized in other languages, especially those that are different from English in terms of word order, syntactic structures, stylistic features and the like. The hypothesis of the present study is that due to the differences between Persian and other Indo-European languages, including English, in those aspects (word order, syntactic structures, stylistic features, etc.), the claimed universal features shown in the other languages are not present in Persian, not at least with the same quality.
Investigation of TUs in Persian has also been largely neglected and faces some drawbacks. Some of these drawbacks are due to research methods (such as manual investigation and not benefiting from corpus investigation tools) and some are related to the limitation of data to only literary texts and novels. Focusing on comparison of source with target text(s) (addressing only S-universals using parallel corpora) and neglecting the examination of T-universals (using comparative corpora) is another limitation of such studies on Persian language. The present study intends to study the salient and distinctive linguistic features of translational Persian using corpus methodology and comparing translated texts with other comparable original writings, thus shedding light on the existence of the claimed universal features in Persian. In this regard, the T-universals of simplification and explicitation were selected and specific linguistic features were examined that can signal the presence of the selected universals. As an instance of non-literary writings, a medium-sized corpus consisting of two sets of expository academic and general humanities texts, namely philosophy and texts about literature (such as general informative texts about literature, review and critique of literary works, etc.), was analysed.
2. Background
The search for TUs dates back to the mid-nineties, where this topic led to a surge of interest among researchers, especially since the emergence of corpora as a research tool in Translation Studies. Searching through the existing literature on TUs shows that the research carried out on S-universals outweigh the studies on T-universals. Among others, simplification and explicitation are the two T-universals investigated in the present study.
A number of studies have been done on simplification as a universal feature in translation at the lexical, syntactic and stylistic level (e.g. see Laviosa-Braithwaite, 1996; Malmkjær, 1997; Laviosa, 1998; Cvrček & Chlumská, 2015). Taking a closer look at these studies highlighted some disagreements. For example, regarding mean sentence length, Laviosa (1998) (English), Xiao and Yue (2009) (Chinese), and Ilisei et al. (2009) (Spanish) showed that mean sentence length in translational language is significantly higher than original writings. But, contrary to these three studies, Malmkjær (1997) believed that stronger punctuations may result in shorter sentences in translated texts. Also, Xiao (2010) and Xiao and Hu (2015) found that sentences in original Chinese are relatively longer than translated texts, although this difference was not significant.
A number of other studies have further demonstrated evidence for explicitation or the tendency in translational language to make explicit what has been implicit in the source text, thus making it different from original writings (Blum-Kulka, 1986; Toury, 1991; Baker, 1996; Øverås, 1998; Olohan & Baker, 2000; Xiao, 2010). Although this feature is found in translations at different lexical, syntactic, and textual levels, "there is variation even in these results, which could be explained in terms of the level of language studied, or the genre of the texts" (Mauranen, 2007, p. 39). There is still a long way to go to determine whether explicitation is a universal feature or not, as most of the data in the literature is based on Western languages, especially English.
Much of the criticism that the topic of TUs has attracted relates to the fact that most studies have only focused on Western languages and failed to move beyond and scrutinize others, a fact that is also reflected in the small body of literature on Persian. What we know about the possible presence of TUs in translational Persian is mostly based on studies that were carried out on S-universals and are limited in one way or another (e.g. Ghamkhah & Khazaee Farid, 2011; Salimi & Askarzadeh Torghabeh, 2015; Vahedi Kia & Ouliaeinia, 2016; Ahangar & Rahnemoon, 2019). In general, these limitations can be classified into seven categories:
direction being restricted to comparison of source with target text(s) (addressing only S-universals)
lack of variety in the source language (almost all studies feature English as the source)
universals (all studies are on the four recurrent features of translation proposed by Baker (1996))
genre (almost all studies focus on literary texts)
size (very small-scale studies, mainly on selected parts of one or two books)
methodology (using manual investigation and not benefiting from corpus investigation tools)
source of data collection (all data were collected from books and published works, ignoring online translated materials available)
3. Corpus Design and Method
In the present study, a comparative corpus was used which includes two subcorpora: original Persian texts and English-Persian translated texts. Each component consisted of one hundred extracts, each of 3000-word length, taken randomly from books and webpages, thus amounting to 300,000 words for each subcorpus and 600,000 words on the whole. The current literature on Persian language has failed to move beyond literary texts. Contrary to the predominance of studies on European languages and small-sized corpus-based works on Persian, all confined to literary texts and books, the data for the present research was collected from books and webpages on two non-literary fields in Humanities, philosophy and texts about literature (such as general informative texts about literature, review and critique of literary works, etc.). Finally, both sides of the corpus are comparable in terms of number of samples, size, genre and sampling period.
After collecting each sample, a header was assigned to it. For samples collected from books, this header contains information about the book title and year of publication, and for websites, it includes the title of the text, date of the post and the webpage URL. To normalize the data, we employed Virastyar, a Persian MS-Word add-in. Moreover, for segmentation, tokenization and POS tagging, we utilized tools developed by Mojgan Seraji (2015) for Persian, namely SeTPer (sentence segmenter and tokenizer) and TagPer (POS tagger). In addition, after analyzing different corpu-analyzer tools (namely WordSmith, AntConc, Sketch Engine, and LancsBox), it was found that the best and most adaptable software for analyzing Persian texts is "WordSmith".
Two universal features of simplification and explicitation were selected for investigation. The presence of universals was identified through a number of features. For simplification, the study used the three signs discussed in Laviosa-Braithwaite (1996) where she concluded that translational language uses lower lexical density, shows less lexical variety, and reports greater mean sentence length. For explicitation, the higher frequency of connectives and cohesive ties in translated than non-translated language was employed (Olohan & Baker, 2000; Chen, 2006).
4. Results
The four different lexical and syntactic features of translational Persian were examined in the corpus under investigation, namely lexical density, lexical variety, mean sentence length, and frequency of connectives. First, regarding simplification, it was found that translational Persian in the comparative corpus used in this study has a higher lexical density, although this difference was minor and was not statistically significant. Also, the lexical variety in translational Persian was greater than non-translational texts. In addition, the study of the mean sentence length showed that sentences in original texts are slightly longer than translated texts. Comparing the two subcorpora, the texts "about literature" show higher lexical density and variety (or richness), and the sentences in philosophical texts were longer, which can be interpreted as field (also genre) variations and their idiosyncratic linguistic features. Finally, regarding explicitation, the total number of connectives was higher in the original texts than in translated texts. However, no clear overall tendency was detected in either subcorpus favoring connectives more than the other. Some connectives were more frequent in translations and some in original texts. Further, some connectives followed no trend as they were more frequent in one field but less frequent in the other.
5. Discussion & Conclusion
Moreover, the data and findings provide further support for the controversies over the strong version of TU hypotheses and raise intriguing questions regarding the presence of universal features in translations as none of the results for the four features addressed were in line with previously proposed T-universals. Therefore, the results of this study support the hypothesis that the claimed universal features, due to linguistic differences, are not present in Persian (at least to the same quality). Contrary to many previous studies (such as the detailed investigation of Ilisei et al. (2009)), features like lower lexical richness and density, greater mean sentence length and higher frequency of connectives might possibly not be among the most salient, universal (at least in its global sense) features indicative of the simplification and explicitation hypotheses. Therefore, it can be cocluded that the findings of this study indicate the specific features of translational (from an English source) and original Persian texts. In general, the present study shows that, in contrast to what might be assumed, simplification and explicitation as so-called translation universals may not be really universal as discussed by Baker (1993) and Eskola (2004), because they are not universally present in all translated texts, at least as far as thid research accounts for translational Persian psychology and sociology.
Whereas a number of studies support simplification and explicitation as translation universals, these linguistic features have been challenged by some other studies, especially when language pairs and genres vary and move from the more investigated languages and genres (Western languages, literary texts) to the less investigated ones (non-Western languages, non-literary texts) (Chesterman, 2004; Mauranen, 2007; Xiao & Hu, 2015). It seems that the assumption of the presence of similar linguistic features in all translations needs to be revised. Therefore, it is better to be cautious in presenting such generalizations and to reclassify them under what Eskola (2004) labels local translation law rather than universal translation law. In fact, it should be noted that some of the theories presented have been formulated using only a pair of specific language pairs or texts, and may not apply to other languages or genres and should therefore be limited and narrowed down. As it was shown in this study, both T-universals examined here were not present in Persian with the same quality as indicated by previous research.
Since it is not possible to proceed with any claim about the presence of universal tendencies in translations without validation, further work needs to be done to establish whether TU hypotheses are supported, at least in their current account, in other, especially unexamined, languages and genres. Although the results of the present study did not support any of the hypotheses presented in the previous studies, this may not be a good reason to dismiss the universals altogether. The authors believe that, instead of abandoning the whole possibility of translations displaying common features, we may find, at least, new tendencies that are different from those of the previous hypotheses; for example, simplification in translational language may be universally manifested through features other than lower lexical density or less lexical variety. Nevertheless, the present study indicated that the claim of the existence of "universal" features in its absolute sense (in all languages and text types) is unfounded. Much more research should be done on translational Persian and other non-European languages in order to clarify the validity and nature of TUs and the role of language, text type, translator skills and other intervening aspects involved in the minifestation of certain linguistic features in translations.
1. Introduction
One of the most significant topics whithin Descriptive Translation Studies (DTS) is Translation Universals (TUs), first clearly articulated by Mona Baker in her work (1993). TU hypotheses are concerned with typical linguistic features that makes translational language different from other linguistic variants. According to Hansen and Teich (2001), “it is commonly assumed in translation studies that translations are specific kinds of texts that are different not only from their original source language (SL) texts, but also from comparable original texts in the same language as the target language (TL)” (p. 44). Such claims can be examined either manually or by means of corpus-based analytical tools. Corpora have been a reliable popular tool among researchers since the convergence of corpus-based empirical methodology and linguistic studies, including Translation Studies, during the 1990s.
Over the last three decades, many studies have been conducted on theories of TUs and evidence of specific linguistic features of translational language has been provided, but almost all of these studies have been on Western languages, especially English. Chesterman (2004) divides the TUs into "S-universals" and "T-universals." The first category refers to "universal differences between translations and their source texts" (ibid, p. 39) and the second category refers to the differences in the linguistic features of translations (target texts) as compared to non-translated, native TL texts. Although some TUs may usually be investigated within one category, some can be examined from the perspective of both groups.
Some interpret the so-called TUs as an inseparable part of any translational language and are in line with the theories presented in the literature of translation studies. It should be noted, however, that if a linguistic feature is to be considered a "universal," it should be found in translations into all languages, but, in fact, almost all the literature on TUs is devoted to research on Western languages, especially English. Only a few examples can be found (e.g. Xiao & Hu, 2015) that have studied universals in non-European languages. In addition, research on S-universals outweighs work on T-universals. Hence, such claims as the existence of ‘universal’ features, in the strict sense of the word ‘universal’, is highly debatable, unless they are scrutinized in other languages, especially those that are different from English in terms of word order, syntactic structures, stylistic features and the like. The hypothesis of the present study is that due to the differences between Persian and other Indo-European languages, including English, in those aspects (word order, syntactic structures, stylistic features, etc.), the claimed universal features shown in the other languages are not present in Persian, not at least with the same quality.
Investigation of TUs in Persian has also been largely neglected and faces some drawbacks. Some of these drawbacks are due to research methods (such as manual investigation and not benefiting from corpus investigation tools) and some are related to the limitation of data to only literary texts and novels. Focusing on comparison of source with target text(s) (addressing only S-universals using parallel corpora) and neglecting the examination of T-universals (using comparative corpora) is another limitation of such studies on Persian language. The present study intends to study the salient and distinctive linguistic features of translational Persian using corpus methodology and comparing translated texts with other comparable original writings, thus shedding light on the existence of the claimed universal features in Persian. In this regard, the T-universals of simplification and explicitation were selected and specific linguistic features were examined that can signal the presence of the selected universals. As an instance of non-literary writings, a medium-sized corpus consisting of two sets of expository academic and general humanities texts, namely philosophy and texts about literature (such as general informative texts about literature, review and critique of literary works, etc.), was analysed.
2. Background
The search for TUs dates back to the mid-nineties, where this topic led to a surge of interest among researchers, especially since the emergence of corpora as a research tool in Translation Studies. Searching through the existing literature on TUs shows that the research carried out on S-universals outweigh the studies on T-universals. Among others, simplification and explicitation are the two T-universals investigated in the present study.
A number of studies have been done on simplification as a universal feature in translation at the lexical, syntactic and stylistic level (e.g. see Laviosa-Braithwaite, 1996; Malmkjær, 1997; Laviosa, 1998; Cvrček & Chlumská, 2015). Taking a closer look at these studies highlighted some disagreements. For example, regarding mean sentence length, Laviosa (1998) (English), Xiao and Yue (2009) (Chinese), and Ilisei et al. (2009) (Spanish) showed that mean sentence length in translational language is significantly higher than original writings. But, contrary to these three studies, Malmkjær (1997) believed that stronger punctuations may result in shorter sentences in translated texts. Also, Xiao (2010) and Xiao and Hu (2015) found that sentences in original Chinese are relatively longer than translated texts, although this difference was not significant.
A number of other studies have further demonstrated evidence for explicitation or the tendency in translational language to make explicit what has been implicit in the source text, thus making it different from original writings (Blum-Kulka, 1986; Toury, 1991; Baker, 1996; Øverås, 1998; Olohan & Baker, 2000; Xiao, 2010). Although this feature is found in translations at different lexical, syntactic, and textual levels, "there is variation even in these results, which could be explained in terms of the level of language studied, or the genre of the texts" (Mauranen, 2007, p. 39). There is still a long way to go to determine whether explicitation is a universal feature or not, as most of the data in the literature is based on Western languages, especially English.
Much of the criticism that the topic of TUs has attracted relates to the fact that most studies have only focused on Western languages and failed to move beyond and scrutinize others, a fact that is also reflected in the small body of literature on Persian. What we know about the possible presence of TUs in translational Persian is mostly based on studies that were carried out on S-universals and are limited in one way or another (e.g. Ghamkhah & Khazaee Farid, 2011; Salimi & Askarzadeh Torghabeh, 2015; Vahedi Kia & Ouliaeinia, 2016; Ahangar & Rahnemoon, 2019). In general, these limitations can be classified into seven categories:
direction being restricted to comparison of source with target text(s) (addressing only S-universals)
lack of variety in the source language (almost all studies feature English as the source)
universals (all studies are on the four recurrent features of translation proposed by Baker (1996))
genre (almost all studies focus on literary texts)
size (very small-scale studies, mainly on selected parts of one or two books)
methodology (using manual investigation and not benefiting from corpus investigation tools)
source of data collection (all data were collected from books and published works, ignoring online translated materials available)
3. Corpus Design and Method
In the present study, a comparative corpus was used which includes two subcorpora: original Persian texts and English-Persian translated texts. Each component consisted of one hundred extracts, each of 3000-word length, taken randomly from books and webpages, thus amounting to 300,000 words for each subcorpus and 600,000 words on the whole. The current literature on Persian language has failed to move beyond literary texts. Contrary to the predominance of studies on European languages and small-sized corpus-based works on Persian, all confined to literary texts and books, the data for the present research was collected from books and webpages on two non-literary fields in Humanities, philosophy and texts about literature (such as general informative texts about literature, review and critique of literary works, etc.). Finally, both sides of the corpus are comparable in terms of number of samples, size, genre and sampling period.
After collecting each sample, a header was assigned to it. For samples collected from books, this header contains information about the book title and year of publication, and for websites, it includes the title of the text, date of the post and the webpage URL. To normalize the data, we employed Virastyar, a Persian MS-Word add-in. Moreover, for segmentation, tokenization and POS tagging, we utilized tools developed by Mojgan Seraji (2015) for Persian, namely SeTPer (sentence segmenter and tokenizer) and TagPer (POS tagger). In addition, after analyzing different corpu-analyzer tools (namely WordSmith, AntConc, Sketch Engine, and LancsBox), it was found that the best and most adaptable software for analyzing Persian texts is "WordSmith".
Two universal features of simplification and explicitation were selected for investigation. The presence of universals was identified through a number of features. For simplification, the study used the three signs discussed in Laviosa-Braithwaite (1996) where she concluded that translational language uses lower lexical density, shows less lexical variety, and reports greater mean sentence length. For explicitation, the higher frequency of connectives and cohesive ties in translated than non-translated language was employed (Olohan & Baker, 2000; Chen, 2006).
4. Results
The four different lexical and syntactic features of translational Persian were examined in the corpus under investigation, namely lexical density, lexical variety, mean sentence length, and frequency of connectives. First, regarding simplification, it was found that translational Persian in the comparative corpus used in this study has a higher lexical density, although this difference was minor and was not statistically significant. Also, the lexical variety in translational Persian was greater than non-translational texts. In addition, the study of the mean sentence length showed that sentences in original texts are slightly longer than translated texts. Comparing the two subcorpora, the texts "about literature" show higher lexical density and variety (or richness), and the sentences in philosophical texts were longer, which can be interpreted as field (also genre) variations and their idiosyncratic linguistic features. Finally, regarding explicitation, the total number of connectives was higher in the original texts than in translated texts. However, no clear overall tendency was detected in either subcorpus favoring connectives more than the other. Some connectives were more frequent in translations and some in original texts. Further, some connectives followed no trend as they were more frequent in one field but less frequent in the other.
5. Discussion & Conclusion
Moreover, the data and findings provide further support for the controversies over the strong version of TU hypotheses and raise intriguing questions regarding the presence of universal features in translations as none of the results for the four features addressed were in line with previously proposed T-universals. Therefore, the results of this study support the hypothesis that the claimed universal features, due to linguistic differences, are not present in Persian (at least to the same quality). Contrary to many previous studies (such as the detailed investigation of Ilisei et al. (2009)), features like lower lexical richness and density, greater mean sentence length and higher frequency of connectives might possibly not be among the most salient, universal (at least in its global sense) features indicative of the simplification and explicitation hypotheses. Therefore, it can be cocluded that the findings of this study indicate the specific features of translational (from an English source) and original Persian texts. In general, the present study shows that, in contrast to what might be assumed, simplification and explicitation as so-called translation universals may not be really universal as discussed by Baker (1993) and Eskola (2004), because they are not universally present in all translated texts, at least as far as thid research accounts for translational Persian psychology and sociology.
Whereas a number of studies support simplification and explicitation as translation universals, these linguistic features have been challenged by some other studies, especially when language pairs and genres vary and move from the more investigated languages and genres (Western languages, literary texts) to the less investigated ones (non-Western languages, non-literary texts) (Chesterman, 2004; Mauranen, 2007; Xiao & Hu, 2015). It seems that the assumption of the presence of similar linguistic features in all translations needs to be revised. Therefore, it is better to be cautious in presenting such generalizations and to reclassify them under what Eskola (2004) labels local translation law rather than universal translation law. In fact, it should be noted that some of the theories presented have been formulated using only a pair of specific language pairs or texts, and may not apply to other languages or genres and should therefore be limited and narrowed down. As it was shown in this study, both T-universals examined here were not present in Persian with the same quality as indicated by previous research.
Since it is not possible to proceed with any claim about the presence of universal tendencies in translations without validation, further work needs to be done to establish whether TU hypotheses are supported, at least in their current account, in other, especially unexamined, languages and genres. Although the results of the present study did not support any of the hypotheses presented in the previous studies, this may not be a good reason to dismiss the universals altogether. The authors believe that, instead of abandoning the whole possibility of translations displaying common features, we may find, at least, new tendencies that are different from those of the previous hypotheses; for example, simplification in translational language may be universally manifested through features other than lower lexical density or less lexical variety. Nevertheless, the present study indicated that the claim of the existence of "universal" features in its absolute sense (in all languages and text types) is unfounded. Much more research should be done on translational Persian and other non-European languages in order to clarify the validity and nature of TUs and the role of language, text type, translator skills and other intervening aspects involved in the minifestation of certain linguistic features in translations.
Send email to the article author
| Rights and permissions | |
 |
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License. |